5.1 Data formats
.h5
diff --git a/img/01_session2/1-example_viewer_img.png b/img/01_session2/1-example_viewer_img.png new file mode 100644 index 0000000..05c3b81 Binary files /dev/null and b/img/01_session2/1-example_viewer_img.png differ diff --git a/interactivity-with-the-rspatial-ecosystem.html b/interactivity-with-the-rspatial-ecosystem.html index b04cb70..ffd8c05 100644 --- a/interactivity-with-the-rspatial-ecosystem.html +++ b/interactivity-with-the-rspatial-ecosystem.html @@ -522,7 +522,7 @@17.1 Kriging
17.1.1 Visium technology
-
+
Figure 17.1: Overview of Visium. Source: 10X Genomics.
@@ -542,94 +542,94 @@
17.1.3 Generating a geojson file
17.2 Downloading the dataset
We first need to import a dataset that we want to perform kriging on.
-data_directory <- "data"
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz")
+data_directory <- "data"
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”.
17.3 Importing visium data
We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network.
-library(Giotto)
-
-save_directory <- 'results'
-visium_save_directory <- file.path(save_directory, 'visium_mouse_brain')
-subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/')
-
-instrs <- createGiottoInstructions(show_plot = TRUE,
- save_plot = TRUE,
- save_dir = visium_save_directory)
-
-v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs)
-
-# Subset to in tissue only
-cm = pDataDT(v_brain)
-in_tissue_barcodes = cm[in_tissue == 1]$cell_ID
-v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes)
-
-# Filter
-v_brain = filterGiotto(gobject = v_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = c('raw'))
-
-# Normalize
-v_brain = normalizeGiotto(gobject = v_brain,
- scalefactor = 6000,
- verbose = TRUE)
-
-# Add stats
-v_brain = addStatistics(gobject = v_brain)
-
-# ID HVF
-v_brain = calculateHVF(gobject = v_brain, method = "cov_loess")
-fm = fDataDT(v_brain)
-hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID
-length(hv_feats)
-
-# Dimension Reductions
-v_brain = runPCA(gobject = v_brain,
- feats_to_use = hv_feats)
-
-v_brain = runUMAP(v_brain,
- dimensions_to_use = 1:10,
- n_neighbors = 15,
- set_seed = TRUE)
-
-# NN Network
-v_brain = createNearestNetwork(gobject = v_brain,
- dimensions_to_use = 1:10,
- k = 15)
-# Leiden Cluster
-v_brain = doLeidenCluster(gobject = v_brain,
- resolution = 0.4,
- n_iterations = 1000,
- set_seed = TRUE)
-
-# Spatial Network (kNN)
-v_brain <- createSpatialNetwork(gobject = v_brain,
- method = 'kNN',
- k = 5,
- maximum_distance_knn = 400,
- name = 'spatial_network')
-
-spatPlot2D(gobject = v_brain,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 1.5,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = T,
- save_param = list(save_name = '03_ses6_1_vis_spat'))
+library(Giotto)
+
+save_directory <- 'results'
+visium_save_directory <- file.path(save_directory, 'visium_mouse_brain')
+subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/')
+
+instrs <- createGiottoInstructions(show_plot = TRUE,
+ save_plot = TRUE,
+ save_dir = visium_save_directory)
+
+v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs)
+
+# Subset to in tissue only
+cm = pDataDT(v_brain)
+in_tissue_barcodes = cm[in_tissue == 1]$cell_ID
+v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes)
+
+# Filter
+v_brain = filterGiotto(gobject = v_brain,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1000,
+ expression_values = c('raw'))
+
+# Normalize
+v_brain = normalizeGiotto(gobject = v_brain,
+ scalefactor = 6000,
+ verbose = TRUE)
+
+# Add stats
+v_brain = addStatistics(gobject = v_brain)
+
+# ID HVF
+v_brain = calculateHVF(gobject = v_brain, method = "cov_loess")
+fm = fDataDT(v_brain)
+hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID
+length(hv_feats)
+
+# Dimension Reductions
+v_brain = runPCA(gobject = v_brain,
+ feats_to_use = hv_feats)
+
+v_brain = runUMAP(v_brain,
+ dimensions_to_use = 1:10,
+ n_neighbors = 15,
+ set_seed = TRUE)
+
+# NN Network
+v_brain = createNearestNetwork(gobject = v_brain,
+ dimensions_to_use = 1:10,
+ k = 15)
+# Leiden Cluster
+v_brain = doLeidenCluster(gobject = v_brain,
+ resolution = 0.4,
+ n_iterations = 1000,
+ set_seed = TRUE)
+
+# Spatial Network (kNN)
+v_brain <- createSpatialNetwork(gobject = v_brain,
+ method = 'kNN',
+ k = 5,
+ maximum_distance_knn = 400,
+ name = 'spatial_network')
+
+spatPlot2D(gobject = v_brain,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 1.5,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = T,
+ save_param = list(save_name = '03_ses6_1_vis_spat'))
Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain.
-
+
Figure 17.2: Mouse brain spatial plot showing leiden clustering
@@ -638,17 +638,17 @@
17.3 Importing visium data
17.3.1 Identifying spatially organized features
We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time.
-# Spatially Variable Features
-ranktest = binSpect(v_brain,
- bin_method = 'rank',
- calc_hub = T,
- hub_min_int = 5,
- spatial_network_name = 'spatial_network',
- do_parallel = T,
- cores = 8) #not able to provide a seed number, so do not set one
-
-# Getting the top 15 spatially organized genes
-ext_spatial_features = ranktest[1:15,]$feats
+# Spatially Variable Features
+ranktest = binSpect(v_brain,
+ bin_method = 'rank',
+ calc_hub = T,
+ hub_min_int = 5,
+ spatial_network_name = 'spatial_network',
+ do_parallel = T,
+ cores = 8) #not able to provide a seed number, so do not set one
+
+# Getting the top 15 spatially organized genes
+ext_spatial_features = ranktest[1:15,]$feats
@@ -656,27 +656,27 @@ 17.4 Adding cell polygons to Giot
17.4.1 Read in the poly information
First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object.
-# Read in the data
-stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson")
-
-stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path,
- name = "stardist_cell",
- calc_centroids = TRUE)
-
-stardist_cell_gpoly <- flip(stardist_cell_gpoly)
+# Read in the data
+stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson")
+
+stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path,
+ name = "stardist_cell",
+ calc_centroids = TRUE)
+
+stardist_cell_gpoly <- flip(stardist_cell_gpoly)
17.4.2 Vizualizing polygons
Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain.
-
-
+
+
Figure 17.3: Mouse brain cell polygons from the visium dataset
-
+
Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped
@@ -686,8 +686,8 @@
17.4.2 Vizualizing polygons
17.4.3 Showing Giotto object prior to polygon addition
Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon.
-
-
+
+
Figure 17.5: Giotto object before adding subcellular polygons.
@@ -697,11 +697,11 @@
17.4.3 Showing Giotto object prio
17.4.4 Adding polygons to giotto object
After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject.
-v_brain <- addGiottoPolygons(v_brain,
- gpolygons = list('stardist_cell' = stardist_cell_gpoly))
-
-print(v_brain)
-
+v_brain <- addGiottoPolygons(v_brain,
+ gpolygons = list('stardist_cell' = stardist_cell_gpoly))
+
+print(v_brain)
+
Figure 17.6: Giotto object after to adding subcellular polygons.
@@ -711,9 +711,9 @@
17.4.4 Adding polygons to giotto
17.4.5 Check polygon information
We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots.
-
-
+
+
Figure 17.7: Polygon information for stardist_cell.
@@ -727,23 +727,23 @@
17.5 Performing kriging17.5.1 Interpolating features
Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be
analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required.
-future::plan(future::multisession()) # comment out for single threading
-subcell_v_brain <- interpolateFeature(v_brain,
- spat_unit = 'cell',
- feat_type = 'rna',
- ext = ext(v_brain),
- feats = ext_spatial_features,
- overwrite = T)
-
-print(subcell_v_brain)
-
+future::plan(future::multisession()) # comment out for single threading
+subcell_v_brain <- interpolateFeature(v_brain,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ ext = ext(v_brain),
+ feats = ext_spatial_features,
+ overwrite = T)
+
+print(subcell_v_brain)
+
Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right).
For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1.
-
+
Figure 17.9: Raster of gene expression interpolation for Pantr1
@@ -754,18 +754,18 @@
17.5.1 Interpolating features17.5.2 Expression overlap
The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier.
This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run.
-subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain,
- name_overlap = "rna",
- spatial_info = "stardist_cell",
- image_names = ext_spatial_features)
-
-subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain,
- poly_info = "stardist_cell",
- feat_info = "rna",
- aggr_function = "sum",
- type='intensity')
+subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain,
+ name_overlap = "rna",
+ spatial_info = "stardist_cell",
+ image_names = ext_spatial_features)
+
+subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain,
+ poly_info = "stardist_cell",
+ feat_info = "rna",
+ aggr_function = "sum",
+ type='intensity')
After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified.
-
+
Figure 17.10: Gene expression for cells based on interpolation.
@@ -776,70 +776,70 @@
17.5.2 Expression overlap
17.6 Reading in larger dataset
For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here.
-
+
17.7 Analyzing interpolated features
17.7.1 Filter and normalization
Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis.
-subcell_v_brain <- filterGiotto(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- expression_values = "raw",
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-
-subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- scalefactor = 6000,
- verbose = T)
+subcell_v_brain <- filterGiotto(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ expression_values = "raw",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+
+subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ scalefactor = 6000,
+ verbose = T)
17.7.2 Visualizing gene expression from interpolated expression
Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data.
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Sparc',
- point_size = 2,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_sparc_vis'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'stardist_cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Sparc',
- point_size = 0.6,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_sparc'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'cell',
- gradient_style = 'sequential',
- feats = 'Pantr1',
- cell_color_gradient = 'Geyser',
- point_size = 2,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_pantr1_vis'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'stardist_cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Pantr1',
- point_size = 0.6,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_pantr1'))
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Sparc',
+ point_size = 2,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_sparc_vis'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Sparc',
+ point_size = 0.6,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_sparc'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'cell',
+ gradient_style = 'sequential',
+ feats = 'Pantr1',
+ cell_color_gradient = 'Geyser',
+ point_size = 2,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_pantr1_vis'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Pantr1',
+ point_size = 0.6,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_pantr1'))
Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer.
-
+
Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom).
@@ -848,37 +848,37 @@
17.7.2 Visualizing gene expressio
17.7.3 Run PCA
-
+
17.7.4 Clustering
-# UMAP
-subcell_v_brain <- runUMAP(subcell_v_brain,
- spat_unit = "stardist_cell",
- dimensions_to_use = 1:15,
- n_neighbors = 1000,
- min_dist = 0.001,
- spread = 1)
-
-# NN Network
-subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- dimensions_to_use = 1:10,
- feats_to_use = hv_feats,
- expression_values = 'normalized',
- k = 70)
-
-subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- resolution = 0.15,
- n_iterations = 100,
- partition_type = 'RBConfigurationVertexPartition')
-
-plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus')
-
+# UMAP
+subcell_v_brain <- runUMAP(subcell_v_brain,
+ spat_unit = "stardist_cell",
+ dimensions_to_use = 1:15,
+ n_neighbors = 1000,
+ min_dist = 0.001,
+ spread = 1)
+
+# NN Network
+subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ dimensions_to_use = 1:10,
+ feats_to_use = hv_feats,
+ expression_values = 'normalized',
+ k = 70)
+
+subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ resolution = 0.15,
+ n_iterations = 100,
+ partition_type = 'RBConfigurationVertexPartition')
+
+plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus')
+
Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering.
@@ -888,27 +888,27 @@
17.7.4 Clustering
17.7.5 Visualizing clustering
Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster.
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.5,
- point_shape = 'no_border',
- background_color = 'black',
- save_plot = F,
- show_legend = TRUE)
-
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat'))
-
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.5,
+ point_shape = 'no_border',
+ background_color = 'black',
+ save_plot = F,
+ show_legend = TRUE)
+
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat'))
+
Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
@@ -918,37 +918,37 @@
17.7.5 Visualizing clustering
17.7.6 Cropping objects
We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below.
-subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
- spat_unit = ":all:",
- x_min = 4000,
- x_max = 7000,
- y_min = -6500,
- y_max = -3500,
- z_max = NULL,
- z_min = NULL)
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 2,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_vis_spat_crop'))
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat_crop'))
-
+subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
+ spat_unit = ":all:",
+ x_min = 4000,
+ x_max = 7000,
+ y_min = -6500,
+ y_max = -3500,
+ z_max = NULL,
+ z_min = NULL)
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 2,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_vis_spat_crop'))
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat_crop'))
+
Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
diff --git a/interoperability-with-isolated-tools.html b/interoperability-with-isolated-tools.html
index ceb2d0a..8cca7e6 100644
--- a/interoperability-with-isolated-tools.html
+++ b/interoperability-with-isolated-tools.html
@@ -526,1521 +526,1521 @@
16.1.1 Dataset downloaddata_path <- "data"
-dir.create(data_path)
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
- destfile = file.path(data_path,"counts.h5ad"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
- destfile = file.path(data_path,"cell_labels.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
+data_path <- "data"
+dir.create(data_path)
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
+ destfile = file.path(data_path,"counts.h5ad"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
+ destfile = file.path(data_path,"cell_labels.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
16.1.2 Create the Giotto object
-library(Giotto)
-library(reticulate)
-
-## Set instructions
-results_folder <- "results"
-dir.create(results_folder)
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-
-## load data
-
-### meta data
-meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-colnames(meta_df)[[1]] <- "cell_ID"
-slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
-slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
-
-### counts matrix
-scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
-ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
-reticulate::source_python(ad2g_path)
-adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
-X <- extract_expression(adata)
-cID <- extract_cell_IDs(adata)
-fID <- extract_feat_IDs(adata)
-rownames(X) <- fID
-colnames(X) <- cID
-slice1_filter <- cID %in% slice1_meta_df$cell_ID
-slice2_filter <- cID %in% slice2_meta_df$cell_ID
-slice1_exp <- X[,slice1_filter]
-slice2_exp <- X[,slice2_filter]
-
-### cell segmentation to cell location
-segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_df <- rbind(segments_1_df, segments_2_df)
-loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
-loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
-loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
-centr.x <- apply(loc.use[,loc.x],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-centr.y <- apply(loc.use[,loc.y],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
-slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
-slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
-
-## giotto obj
-giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
- spatial_locs = slice1_spatial_locs_df,
- cell_metadata = slice1_meta_df)
-giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
- spatial_locs = slice2_spatial_locs_df,
- cell_metadata = slice2_meta_df)
-giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
- gobject_names = c('mouse2_slice229',
- 'mouse2_slice300'), # name for each samples
- join_method = 'z_stack')
-# We skip the processing process here to save time and use the given cell type
-# annotation directly
-
-ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
- cell_type = "subclass",
- output_path = data_path,
- spat_unit = "cell",
- feat_type = "rna",
- verbose = TRUE)
-
-# Cell_ID Sample x y Cell_Type
-# <char> <char> <dbl> <dbl> <char>
-# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
-# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
-# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
-# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
-# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
-# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
+library(Giotto)
+library(reticulate)
+
+## Set instructions
+results_folder <- "results"
+dir.create(results_folder)
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+
+## load data
+
+### meta data
+meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+colnames(meta_df)[[1]] <- "cell_ID"
+slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
+slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
+
+### counts matrix
+scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
+ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
+reticulate::source_python(ad2g_path)
+adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
+X <- extract_expression(adata)
+cID <- extract_cell_IDs(adata)
+fID <- extract_feat_IDs(adata)
+rownames(X) <- fID
+colnames(X) <- cID
+slice1_filter <- cID %in% slice1_meta_df$cell_ID
+slice2_filter <- cID %in% slice2_meta_df$cell_ID
+slice1_exp <- X[,slice1_filter]
+slice2_exp <- X[,slice2_filter]
+
+### cell segmentation to cell location
+segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_df <- rbind(segments_1_df, segments_2_df)
+loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
+loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
+loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
+centr.x <- apply(loc.use[,loc.x],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+centr.y <- apply(loc.use[,loc.y],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
+slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
+slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
+
+## giotto obj
+giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
+ spatial_locs = slice1_spatial_locs_df,
+ cell_metadata = slice1_meta_df)
+giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
+ spatial_locs = slice2_spatial_locs_df,
+ cell_metadata = slice2_meta_df)
+giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
+ gobject_names = c('mouse2_slice229',
+ 'mouse2_slice300'), # name for each samples
+ join_method = 'z_stack')
+# We skip the processing process here to save time and use the given cell type
+# annotation directly
+
+ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
+ cell_type = "subclass",
+ output_path = data_path,
+ spat_unit = "cell",
+ feat_type = "rna",
+ verbose = TRUE)
+
+# Cell_ID Sample x y Cell_Type
+# <char> <char> <dbl> <dbl> <char>
+# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
+# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
+# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
+# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
+# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
+# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
Spatial cell type distributions
-spatPlot2D(giotto_obj,
- group_by = "slice_id",
- cell_color = "subclass",
- point_size = 1,
- point_border_stroke = NA,
- legend_text = 6)
+spatPlot2D(giotto_obj,
+ group_by = "slice_id",
+ cell_color = "subclass",
+ point_size = 1,
+ point_border_stroke = NA,
+ legend_text = 6)
16.1.2.1 Installation ONTraC
You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node.
It will run for less than 10 minutes on this example dataset.
For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time.
-source ~/.bash_profile
-conda create -y -n ONTraC python=3.11
-conda activate ONTraC
-pip install git+https://github.com/gyuanlab/ONTraC.git@V2
-# Retrieving notices: ...working... done
-# Channels:
-# - conda-forge
-# - bioconda
-# - r
-# - pytorch
-# - defaults
-# Platform: osx-arm64
-# Collecting package metadata (repodata.json): ...working... done
-# Solving environment: ...working... done
-#
-# ## Package Plan ##
-#
-# environment location: /Users/wwang/miniconda3/envs/ONTraC
-#
-# added / updated specs:
-# - python=3.11
-#
-#
-# The following packages will be downloaded:
-#
-# package | build
-# ---------------------------|-----------------
-# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
-# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
-# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
-# ------------------------------------------------------------
-# Total: 4.3 MB
-#
-# The following NEW packages will be INSTALLED:
-#
-# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
-# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
-# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
-# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
-# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
-# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
-# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
-# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
-# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
-# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
-# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
-# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
-# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
-# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
-# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
-# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
-#
-#
-#
-# Downloading and Extracting Packages: ...working... done
-# Preparing transaction: ...working... done
-# Verifying transaction: ...working... done
-# Executing transaction: ...working... done
-# #
-# # To activate this environment, use
-# #
-# # $ conda activate ONTraC
-# #
-# # To deactivate an active environment, use
-# #
-# # $ conda deactivate
-#
-# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
-# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
-# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
-# Running command git checkout -b V2 --track origin/V2
-# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
-# Installing build dependencies: started
-# Switched to a new branch 'V2'
-# branch 'V2' set up to track 'origin/V2'.
-# Installing build dependencies: finished with status 'done'
-# Getting requirements to build wheel: started
-# Getting requirements to build wheel: finished with status 'done'
-# Preparing metadata (pyproject.toml): started
-# Preparing metadata (pyproject.toml): finished with status 'done'
-# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
-# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
-# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
-# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
-# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
-# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
-# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
-# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
-# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
-# Collecting session-info (from ONTraC==2.0a9.dev7)
-# Using cached session_info-1.0.0-py3-none-any.whl
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
-# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
-# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
-# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
-# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
-# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
-# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
-# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
-# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
-# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
-# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
-# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
-# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
-# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
-# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
-# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
-# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
-# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
-# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
-# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
-# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
-# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
-# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
-# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
-# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
-# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
-# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
-# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
-# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
-# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
-# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
-# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
-# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
-# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
-# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
-# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
-# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
-# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
-# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
-# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
-# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
-# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
-# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
-# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
-# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
-# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
-# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
-# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
-# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
-# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
-# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
-# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
-# Using cached idna-3.7-py3-none-any.whl (66 kB)
-# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
-# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
-# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
-# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
-# Building wheels for collected packages: ONTraC
-# Building wheel for ONTraC (pyproject.toml): started
-# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
-# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
-# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
-# Successfully built ONTraC
-# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
-# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
+source ~/.bash_profile
+conda create -y -n ONTraC python=3.11
+conda activate ONTraC
+pip install git+https://github.com/gyuanlab/ONTraC.git@V2
+# Retrieving notices: ...working... done
+# Channels:
+# - conda-forge
+# - bioconda
+# - r
+# - pytorch
+# - defaults
+# Platform: osx-arm64
+# Collecting package metadata (repodata.json): ...working... done
+# Solving environment: ...working... done
+#
+# ## Package Plan ##
+#
+# environment location: /Users/wwang/miniconda3/envs/ONTraC
+#
+# added / updated specs:
+# - python=3.11
+#
+#
+# The following packages will be downloaded:
+#
+# package | build
+# ---------------------------|-----------------
+# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
+# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
+# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
+# ------------------------------------------------------------
+# Total: 4.3 MB
+#
+# The following NEW packages will be INSTALLED:
+#
+# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
+# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
+# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
+# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
+# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
+# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
+# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
+# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
+# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
+# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
+# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
+# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
+# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
+# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
+# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
+# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
+#
+#
+#
+# Downloading and Extracting Packages: ...working... done
+# Preparing transaction: ...working... done
+# Verifying transaction: ...working... done
+# Executing transaction: ...working... done
+# #
+# # To activate this environment, use
+# #
+# # $ conda activate ONTraC
+# #
+# # To deactivate an active environment, use
+# #
+# # $ conda deactivate
+#
+# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
+# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
+# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
+# Running command git checkout -b V2 --track origin/V2
+# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
+# Installing build dependencies: started
+# Switched to a new branch 'V2'
+# branch 'V2' set up to track 'origin/V2'.
+# Installing build dependencies: finished with status 'done'
+# Getting requirements to build wheel: started
+# Getting requirements to build wheel: finished with status 'done'
+# Preparing metadata (pyproject.toml): started
+# Preparing metadata (pyproject.toml): finished with status 'done'
+# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
+# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
+# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
+# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
+# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
+# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
+# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
+# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
+# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
+# Collecting session-info (from ONTraC==2.0a9.dev7)
+# Using cached session_info-1.0.0-py3-none-any.whl
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
+# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
+# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
+# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
+# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
+# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
+# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
+# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
+# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
+# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
+# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
+# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
+# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
+# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
+# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
+# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
+# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
+# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
+# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
+# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
+# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
+# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
+# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
+# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
+# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
+# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
+# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
+# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
+# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
+# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
+# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
+# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
+# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
+# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
+# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
+# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
+# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
+# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
+# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
+# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
+# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
+# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
+# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
+# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
+# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
+# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
+# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
+# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
+# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
+# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
+# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
+# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
+# Using cached idna-3.7-py3-none-any.whl (66 kB)
+# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
+# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
+# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
+# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
+# Building wheels for collected packages: ONTraC
+# Building wheel for ONTraC (pyproject.toml): started
+# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
+# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
+# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
+# Successfully built ONTraC
+# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
+# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
16.1.2.2 Running ONTraC
-
-source ~/.bash_profile
-conda activate ONTraC_v2_py311
-ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
-# ##################################################################################
-#
-# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
-# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
-# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
-# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
-# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
-#
-# version: 2.0a11
-#
-# ##################################################################################
-# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
-# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
-# 11:33:23 --- INFO: -------- I/O options -------
-# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
-# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
-# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
-# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
-# 11:33:23 --- INFO: -------- preprocessing options -------
-# 11:33:23 --- INFO: resolution: 10.0
-# 11:33:23 --- INFO: -------- niche net constr options -------
-# 11:33:23 --- INFO: n_cpu: 4
-# 11:33:23 --- INFO: n_neighbors: 50
-# 11:33:23 --- INFO: n_local: 20
-# 11:33:23 --- INFO: embedding_adjust: False
-# 11:33:23 --- INFO: sigma: 1
-# 11:33:23 --- INFO: -------- train options -------
-# 11:33:23 --- INFO: device: cpu
-# 11:33:23 --- INFO: epochs: 1000
-# 11:33:23 --- INFO: batch_size: 0
-# 11:33:23 --- INFO: patience: 100
-# 11:33:23 --- INFO: min_delta: 0.001
-# 11:33:23 --- INFO: min_epochs: 50
-# 11:33:23 --- INFO: seed: 42
-# 11:33:23 --- INFO: lr: 0.03
-# 11:33:23 --- INFO: hidden_feats: 4
-# 11:33:23 --- INFO: k: 6
-# 11:33:23 --- INFO: modularity_loss_weight: 1.0
-# 11:33:23 --- INFO: purity_loss_weight: 100.0
-# 11:33:23 --- INFO: regularization_loss_weight: 0.1
-# 11:33:23 --- INFO: beta: 0.3
-# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
-# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
-# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
-# 11:33:23 --- INFO: ------------- Niche network construct ---------------
-# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
-# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
-# 11:33:28 --- INFO: Generating samples.yaml file.
-# 11:33:28 --- INFO: ------------ Niche network construct end ------------
-# 11:33:28 --- INFO: ------------------------ GNN ------------------------
-# 11:33:28 --- INFO: Loading dataset.
-# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
-# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
-# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
-# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
-# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
-# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
-# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
-# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
-# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
-# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
-# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
-# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
-# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
-# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
-# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
-# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
-# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
-# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
-# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
-# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
-# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
-# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
-# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
-# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
-# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
-# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
-# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
-# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
-# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
-# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
-# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
-# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
-# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
-# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
-# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
-# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
-# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
-# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
-# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
-# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
-# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
-# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
-# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
-# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
-# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
-# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
-# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
-# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
-# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
-# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
-# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
-# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
-# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
-# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
-# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
-# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
-# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
-# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
-# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
-# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
-# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
-# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
-# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
-# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
-# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
-# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
-# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
-# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
-# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
-# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
-# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
-# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
-# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
-# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
-# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
-# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
-# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
-# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
-# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
-# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
-# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
-# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
-# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
-# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
-# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
-# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
-# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
-# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
-# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
-# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
-# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
-# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
-# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
-# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
-# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
-# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
-# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
-# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
-# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
-# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
-# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
-# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
-# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
-# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
-# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
-# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
-# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
-# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
-# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
-# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
-# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
-# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
-# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
-# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
-# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
-# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
-# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
-# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
-# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
-# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
-# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
-# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
-# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
-# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
-# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
-# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
-# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
-# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
-# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
-# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
-# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
-# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
-# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
-# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
-# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
-# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
-# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
-# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
-# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
-# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
-# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
-# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
-# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
-# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
-# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
-# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
-# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
-# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
-# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
-# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
-# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
-# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
-# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
-# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
-# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
-# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
-# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
-# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
-# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
-# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
-# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
-# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
-# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
-# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
-# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
-# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
-# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
-# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
-# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
-# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
-# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
-# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
-# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
-# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
-# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
-# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
-# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
-# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
-# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
-# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
-# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
-# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
-# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
-# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
-# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
-# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
-# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
-# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
-# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
-# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
-# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
-# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
-# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
-# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
-# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
-# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
-# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
-# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
-# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
-# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
-# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
-# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
-# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
-# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
-# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
-# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
-# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
-# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
-# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
-# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
-# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
-# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
-# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
-# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
-# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
-# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
-# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
-# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
-# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
-# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
-# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
-# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
-# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
-# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
-# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
-# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
-# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
-# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
-# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
-# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
-# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
-# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
-# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
-# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
-# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
-# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
-# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
-# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
-# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
-# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
-# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
-# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
-# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
-# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
-# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
-# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
-# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
-# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
-# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
-# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
-# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
-# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
-# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
-# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
-# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
-# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
-# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
-# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
-# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
-# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
-# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
-# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
-# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
-# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
-# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
-# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
-# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
-# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
-# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
-# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
-# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
-# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
-# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
-# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
-# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
-# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
-# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
-# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
-# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
-# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
-# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
-# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
-# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
-# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
-# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
-# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
-# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
-# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
-# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
-# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
-# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
-# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
-# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
-# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
-# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
-# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
-# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
-# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
-# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
-# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
-# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
-# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
-# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
-# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
-# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
-# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
-# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
-# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
-# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
-# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
-# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
-# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
-# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
-# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
-# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
-# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
-# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
-# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
-# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
-# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
-# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
-# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
-# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
-# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
-# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
-# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
-# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
-# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
-# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
-# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
-# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
-# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
-# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
-# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
-# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
-# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
-# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
-# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
-# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
-# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
-# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
-# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
-# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
-# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
-# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
-# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
-# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
-# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
-# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
-# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
-# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
-# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
-# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
-# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
-# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
-# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
-# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
-# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
-# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
-# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
-# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
-# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
-# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
-# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
-# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
-# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
-# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
-# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
-# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
-# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
-# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
-# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
-# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
-# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
-# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
-# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
-# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
-# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
-# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
-# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
-# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
-# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
-# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
-# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
-# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
-# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
-# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
-# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
-# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
-# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
-# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
-# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
-# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
-# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
-# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
-# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
-# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
-# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
-# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
-# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
-# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
-# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
-# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
-# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
-# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
-# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
-# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
-# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
-# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
-# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
-# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
-# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
-# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
-# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
-# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
-# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
-# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
-# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
-# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
-# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
-# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
-# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
-# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
-# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
-# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
-# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
-# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
-# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
-# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
-# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
-# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
-# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
-# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
-# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
-# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
-# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
-# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
-# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
-# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
-# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
-# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
-# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
-# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
-# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
-# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
-# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
-# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
-# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
-# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
-# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
-# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
-# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
-# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
-# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
-# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
-# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
-# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
-# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
-# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
-# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
-# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
-# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
-# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
-# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
-# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
-# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
-# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
-# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
-# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
-# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
-# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
-# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
-# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
-# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
-# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
-# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
-# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
-# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
-# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
-# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
-# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
-# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
-# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
-# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
-# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
-# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
-# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
-# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
-# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
-# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
-# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
-# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
-# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
-# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
-# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
-# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
-# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
-# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
-# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
-# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
-# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
-# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
-# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
-# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
-# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
-# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
-# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
-# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
-# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
-# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
-# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
-# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
-# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
-# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
-# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
-# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
-# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
-# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
-# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
-# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
-# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
-# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
-# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
-# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
-# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
-# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
-# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
-# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
-# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
-# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
-# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
-# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
-# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
-# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
-# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
-# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
-# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
-# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
-# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
-# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
-# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
-# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
-# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
-# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
-# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
-# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
-# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
-# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
-# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
-# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
-# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
-# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
-# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
-# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
-# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
-# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
-# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
-# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
-# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
-# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
-# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
-# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
-# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
-# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
-# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
-# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
-# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
-# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
-# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
-# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
-# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
-# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
-# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
-# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
-# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
-# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
-# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
-# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
-# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
-# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
-# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
-# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
-# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
-# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
-# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
-# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
-# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
-# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
-# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
-# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
-# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
-# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
-# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
-# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
-# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
-# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
-# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
-# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
-# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
-# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
-# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
-# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
-# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
-# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
-# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
-# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
-# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
-# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
-# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
-# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
-# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
-# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
-# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
-# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
-# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
-# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
-# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
-# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
-# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
-# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
-# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
-# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
-# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
-# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
-# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
-# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
-# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
-# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
-# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
-# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
-# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
-# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
-# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
-# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
-# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
-# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
-# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
-# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
-# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
-# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
-# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
-# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
-# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
-# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
-# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
-# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
-# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
-# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
-# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
-# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
-# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
-# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
-# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
-# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
-# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
-# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
-# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
-# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
-# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
-# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
-# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
-# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
-# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
-# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
-# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
-# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
-# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
-# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
-# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
-# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
-# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
-# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
-# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
-# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
-# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
-# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
-# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
-# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
-# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
-# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
-# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
-# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
-# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
-# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
-# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
-# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
-# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
-# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
-# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
-# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
-# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
-# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
-# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
-# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
-# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
-# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
-# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
-# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
-# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
-# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
-# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
-# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
-# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
-# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
-# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
-# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
-# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
-# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
-# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
-# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
-# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
-# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
-# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
-# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
-# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
-# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
-# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
-# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
-# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
-# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
-# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
-# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
-# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
-# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
-# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
-# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
-# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
-# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
-# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
-# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
-# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
-# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
-# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
-# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
-# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
-# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
-# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
-# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
-# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
-# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
-# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
-# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
-# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
-# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
-# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
-# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
-# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
-# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
-# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
-# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
-# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
-# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
-# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
-# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
-# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
-# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
-# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
-# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
-# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
-# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
-# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
-# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
-# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
-# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
-# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
-# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
-# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
-# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
-# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
-# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
-# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
-# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
-# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
-# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
-# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
-# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
-# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
-# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
-# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
-# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
-# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
-# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
-# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
-# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
-# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
-# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
-# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
-# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
-# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
-# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
-# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
-# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
-# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
-# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
-# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
-# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
-# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
-# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
-# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
-# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
-# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
-# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
-# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
-# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
-# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
-# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
-# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
-# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
-# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
-# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
-# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
-# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
-# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
-# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
-# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
-# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
-# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
-# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
-# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
-# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
-# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
-# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
-# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
-# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
-# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
-# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
-# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
-# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
-# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
-# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
-# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
-# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
-# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
-# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
-# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
-# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
-# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
-# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
-# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
-# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
-# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
-# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
-# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
-# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
-# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
-# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
-# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
-# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
-# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
-# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
-# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
-# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
-# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
-# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
-# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
-# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
-# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
-# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
-# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
-# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
-# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
-# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
-# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
-# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
-# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
-# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
-# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
-# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
-# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
-# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
-# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
-# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
-# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
-# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
-# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
-# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
-# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
-# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
-# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
-# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
-# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
-# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
-# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
-# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
-# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
-# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
-# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
-# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
-# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
-# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
-# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
-# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
-# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
-# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
-# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
-# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
-# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
-# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
-# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
-# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
-# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
-# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
-# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
-# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
-# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
-# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
-# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
-# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
-# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
-# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
-# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
-# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
-# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
-# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
-# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
-# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
-# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
-# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
-# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
-# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
-# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
-# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
-# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
-# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
-# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
-# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
-# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
-# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
-# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
-# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
-# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
-# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
-# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
-# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
-# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
-# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
-# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
-# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
-# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
-# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
-# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
-# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
-# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
-# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
-# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
-# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
-# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
-# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
-# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
-# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
-# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
-# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
-# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
-# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
-# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
-# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
-# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
-# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
-# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
-# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
-# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
-# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
-# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
-# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
-# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
-# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
-# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
-# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
-# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
-# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
-# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
-# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
-# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
-# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
-# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
-# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
-# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
-# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
-# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
-# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
-# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
-# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
-# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
-# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
-# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
-# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
-# 11:38:37 --- INFO: Training process end.
-# 11:38:37 --- INFO: Evaluating process start.
-# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
-# 11:38:37 --- INFO: Evaluating process end.
-# 11:38:37 --- INFO: Predicting process start.
-# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
-# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
-# 11:38:38 --- INFO: Predicting process end.
-# 11:38:38 --- INFO: --------------------- GNN end ----------------------
-# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
-# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
-# 11:38:38 --- INFO: Loading dataset.
-# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:38:39 --- INFO: Calculating NTScore for each niche.
-# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
-# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
-# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
-# 11:38:39 --- INFO: Output NTScore tables.
-# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
+
+source ~/.bash_profile
+conda activate ONTraC_v2_py311
+ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
+# ##################################################################################
+#
+# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
+# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
+# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
+# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
+# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
+#
+# version: 2.0a11
+#
+# ##################################################################################
+# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
+# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
+# 11:33:23 --- INFO: -------- I/O options -------
+# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
+# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
+# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
+# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
+# 11:33:23 --- INFO: -------- preprocessing options -------
+# 11:33:23 --- INFO: resolution: 10.0
+# 11:33:23 --- INFO: -------- niche net constr options -------
+# 11:33:23 --- INFO: n_cpu: 4
+# 11:33:23 --- INFO: n_neighbors: 50
+# 11:33:23 --- INFO: n_local: 20
+# 11:33:23 --- INFO: embedding_adjust: False
+# 11:33:23 --- INFO: sigma: 1
+# 11:33:23 --- INFO: -------- train options -------
+# 11:33:23 --- INFO: device: cpu
+# 11:33:23 --- INFO: epochs: 1000
+# 11:33:23 --- INFO: batch_size: 0
+# 11:33:23 --- INFO: patience: 100
+# 11:33:23 --- INFO: min_delta: 0.001
+# 11:33:23 --- INFO: min_epochs: 50
+# 11:33:23 --- INFO: seed: 42
+# 11:33:23 --- INFO: lr: 0.03
+# 11:33:23 --- INFO: hidden_feats: 4
+# 11:33:23 --- INFO: k: 6
+# 11:33:23 --- INFO: modularity_loss_weight: 1.0
+# 11:33:23 --- INFO: purity_loss_weight: 100.0
+# 11:33:23 --- INFO: regularization_loss_weight: 0.1
+# 11:33:23 --- INFO: beta: 0.3
+# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
+# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
+# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
+# 11:33:23 --- INFO: ------------- Niche network construct ---------------
+# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
+# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
+# 11:33:28 --- INFO: Generating samples.yaml file.
+# 11:33:28 --- INFO: ------------ Niche network construct end ------------
+# 11:33:28 --- INFO: ------------------------ GNN ------------------------
+# 11:33:28 --- INFO: Loading dataset.
+# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
+# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
+# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
+# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
+# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
+# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
+# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
+# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
+# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
+# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
+# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
+# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
+# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
+# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
+# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
+# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
+# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
+# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
+# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
+# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
+# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
+# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
+# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
+# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
+# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
+# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
+# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
+# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
+# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
+# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
+# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
+# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
+# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
+# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
+# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
+# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
+# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
+# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
+# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
+# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
+# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
+# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
+# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
+# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
+# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
+# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
+# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
+# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
+# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
+# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
+# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
+# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
+# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
+# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
+# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
+# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
+# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
+# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
+# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
+# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
+# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
+# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
+# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
+# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
+# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
+# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
+# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
+# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
+# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
+# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
+# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
+# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
+# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
+# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
+# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
+# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
+# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
+# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
+# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
+# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
+# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
+# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
+# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
+# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
+# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
+# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
+# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
+# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
+# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
+# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
+# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
+# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
+# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
+# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
+# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
+# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
+# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
+# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
+# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
+# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
+# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
+# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
+# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
+# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
+# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
+# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
+# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
+# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
+# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
+# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
+# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
+# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
+# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
+# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
+# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
+# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
+# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
+# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
+# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
+# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
+# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
+# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
+# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
+# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
+# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
+# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
+# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
+# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
+# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
+# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
+# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
+# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
+# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
+# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
+# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
+# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
+# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
+# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
+# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
+# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
+# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
+# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
+# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
+# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
+# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
+# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
+# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
+# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
+# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
+# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
+# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
+# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
+# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
+# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
+# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
+# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
+# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
+# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
+# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
+# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
+# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
+# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
+# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
+# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
+# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
+# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
+# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
+# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
+# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
+# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
+# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
+# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
+# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
+# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
+# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
+# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
+# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
+# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
+# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
+# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
+# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
+# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
+# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
+# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
+# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
+# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
+# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
+# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
+# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
+# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
+# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
+# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
+# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
+# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
+# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
+# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
+# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
+# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
+# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
+# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
+# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
+# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
+# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
+# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
+# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
+# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
+# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
+# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
+# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
+# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
+# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
+# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
+# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
+# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
+# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
+# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
+# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
+# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
+# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
+# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
+# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
+# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
+# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
+# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
+# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
+# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
+# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
+# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
+# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
+# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
+# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
+# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
+# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
+# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
+# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
+# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
+# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
+# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
+# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
+# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
+# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
+# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
+# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
+# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
+# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
+# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
+# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
+# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
+# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
+# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
+# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
+# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
+# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
+# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
+# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
+# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
+# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
+# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
+# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
+# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
+# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
+# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
+# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
+# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
+# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
+# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
+# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
+# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
+# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
+# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
+# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
+# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
+# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
+# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
+# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
+# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
+# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
+# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
+# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
+# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
+# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
+# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
+# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
+# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
+# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
+# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
+# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
+# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
+# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
+# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
+# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
+# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
+# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
+# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
+# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
+# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
+# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
+# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
+# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
+# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
+# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
+# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
+# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
+# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
+# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
+# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
+# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
+# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
+# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
+# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
+# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
+# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
+# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
+# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
+# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
+# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
+# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
+# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
+# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
+# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
+# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
+# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
+# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
+# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
+# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
+# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
+# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
+# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
+# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
+# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
+# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
+# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
+# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
+# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
+# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
+# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
+# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
+# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
+# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
+# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
+# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
+# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
+# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
+# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
+# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
+# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
+# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
+# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
+# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
+# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
+# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
+# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
+# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
+# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
+# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
+# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
+# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
+# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
+# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
+# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
+# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
+# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
+# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
+# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
+# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
+# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
+# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
+# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
+# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
+# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
+# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
+# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
+# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
+# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
+# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
+# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
+# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
+# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
+# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
+# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
+# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
+# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
+# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
+# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
+# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
+# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
+# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
+# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
+# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
+# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
+# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
+# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
+# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
+# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
+# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
+# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
+# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
+# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
+# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
+# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
+# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
+# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
+# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
+# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
+# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
+# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
+# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
+# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
+# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
+# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
+# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
+# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
+# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
+# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
+# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
+# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
+# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
+# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
+# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
+# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
+# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
+# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
+# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
+# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
+# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
+# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
+# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
+# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
+# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
+# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
+# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
+# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
+# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
+# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
+# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
+# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
+# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
+# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
+# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
+# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
+# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
+# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
+# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
+# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
+# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
+# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
+# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
+# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
+# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
+# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
+# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
+# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
+# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
+# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
+# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
+# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
+# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
+# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
+# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
+# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
+# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
+# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
+# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
+# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
+# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
+# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
+# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
+# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
+# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
+# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
+# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
+# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
+# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
+# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
+# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
+# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
+# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
+# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
+# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
+# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
+# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
+# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
+# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
+# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
+# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
+# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
+# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
+# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
+# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
+# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
+# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
+# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
+# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
+# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
+# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
+# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
+# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
+# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
+# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
+# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
+# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
+# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
+# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
+# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
+# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
+# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
+# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
+# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
+# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
+# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
+# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
+# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
+# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
+# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
+# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
+# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
+# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
+# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
+# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
+# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
+# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
+# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
+# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
+# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
+# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
+# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
+# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
+# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
+# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
+# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
+# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
+# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
+# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
+# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
+# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
+# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
+# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
+# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
+# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
+# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
+# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
+# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
+# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
+# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
+# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
+# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
+# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
+# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
+# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
+# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
+# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
+# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
+# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
+# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
+# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
+# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
+# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
+# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
+# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
+# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
+# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
+# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
+# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
+# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
+# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
+# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
+# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
+# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
+# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
+# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
+# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
+# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
+# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
+# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
+# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
+# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
+# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
+# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
+# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
+# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
+# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
+# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
+# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
+# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
+# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
+# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
+# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
+# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
+# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
+# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
+# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
+# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
+# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
+# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
+# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
+# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
+# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
+# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
+# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
+# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
+# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
+# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
+# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
+# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
+# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
+# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
+# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
+# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
+# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
+# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
+# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
+# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
+# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
+# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
+# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
+# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
+# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
+# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
+# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
+# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
+# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
+# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
+# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
+# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
+# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
+# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
+# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
+# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
+# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
+# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
+# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
+# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
+# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
+# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
+# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
+# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
+# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
+# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
+# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
+# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
+# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
+# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
+# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
+# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
+# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
+# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
+# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
+# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
+# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
+# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
+# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
+# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
+# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
+# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
+# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
+# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
+# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
+# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
+# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
+# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
+# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
+# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
+# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
+# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
+# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
+# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
+# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
+# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
+# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
+# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
+# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
+# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
+# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
+# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
+# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
+# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
+# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
+# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
+# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
+# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
+# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
+# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
+# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
+# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
+# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
+# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
+# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
+# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
+# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
+# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
+# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
+# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
+# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
+# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
+# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
+# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
+# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
+# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
+# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
+# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
+# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
+# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
+# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
+# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
+# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
+# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
+# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
+# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
+# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
+# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
+# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
+# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
+# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
+# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
+# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
+# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
+# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
+# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
+# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
+# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
+# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
+# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
+# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
+# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
+# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
+# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
+# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
+# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
+# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
+# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
+# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
+# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
+# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
+# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
+# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
+# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
+# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
+# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
+# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
+# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
+# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
+# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
+# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
+# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
+# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
+# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
+# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
+# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
+# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
+# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
+# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
+# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
+# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
+# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
+# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
+# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
+# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
+# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
+# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
+# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
+# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
+# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
+# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
+# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
+# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
+# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
+# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
+# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
+# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
+# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
+# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
+# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
+# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
+# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
+# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
+# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
+# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
+# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
+# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
+# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
+# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
+# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
+# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
+# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
+# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
+# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
+# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
+# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
+# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
+# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
+# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
+# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
+# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
+# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
+# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
+# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
+# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
+# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
+# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
+# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
+# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
+# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
+# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
+# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
+# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
+# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
+# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
+# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
+# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
+# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
+# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
+# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
+# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
+# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
+# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
+# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
+# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
+# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
+# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
+# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
+# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
+# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
+# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
+# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
+# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
+# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
+# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
+# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
+# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
+# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
+# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
+# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
+# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
+# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
+# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
+# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
+# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
+# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
+# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
+# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
+# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
+# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
+# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
+# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
+# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
+# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
+# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
+# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
+# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
+# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
+# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
+# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
+# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
+# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
+# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
+# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
+# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
+# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
+# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
+# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
+# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
+# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
+# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
+# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
+# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
+# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
+# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
+# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
+# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
+# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
+# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
+# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
+# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
+# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
+# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
+# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
+# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
+# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
+# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
+# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
+# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
+# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
+# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
+# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
+# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
+# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
+# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
+# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
+# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
+# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
+# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
+# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
+# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
+# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
+# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
+# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
+# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
+# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
+# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
+# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
+# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
+# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
+# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
+# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
+# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
+# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
+# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
+# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
+# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
+# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
+# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
+# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
+# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
+# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
+# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
+# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
+# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
+# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
+# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
+# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
+# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
+# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
+# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
+# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
+# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
+# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
+# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
+# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
+# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
+# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
+# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
+# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
+# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
+# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
+# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
+# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
+# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
+# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
+# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
+# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
+# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
+# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
+# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
+# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
+# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
+# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
+# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
+# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
+# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
+# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
+# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
+# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
+# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
+# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
+# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
+# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
+# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
+# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
+# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
+# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
+# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
+# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
+# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
+# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
+# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
+# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
+# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
+# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
+# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
+# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
+# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
+# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
+# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
+# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
+# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
+# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
+# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
+# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
+# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
+# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
+# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
+# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
+# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
+# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
+# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
+# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
+# 11:38:37 --- INFO: Training process end.
+# 11:38:37 --- INFO: Evaluating process start.
+# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
+# 11:38:37 --- INFO: Evaluating process end.
+# 11:38:37 --- INFO: Predicting process start.
+# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
+# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
+# 11:38:38 --- INFO: Predicting process end.
+# 11:38:38 --- INFO: --------------------- GNN end ----------------------
+# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
+# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
+# 11:38:38 --- INFO: Loading dataset.
+# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:38:39 --- INFO: Calculating NTScore for each niche.
+# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
+# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
+# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
+# 11:38:39 --- INFO: Output NTScore tables.
+# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
16.1.3 visualization
16.1.3.1 load ONTraC results
-
+
The NTScore and binarized niche cluster info were stored in cell metadata
-
-# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
-# <char> <char> <char> <char> <char> <char> <char> <int> <num>
-# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
-# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
-# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
-# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
-# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
-# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
+
+# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
+# <char> <char> <char> <char> <char> <char> <char> <int> <num>
+# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
+# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
+# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
+# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
+# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
+# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
The probability matrix of each cell assigned to each niche cluster and
connectivity between niche cluster were stored here.
-
-
+
+
16.1.3.2 niche cluster prob
-spatFeatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'niche cluster',
- expression_values = "prob",
- group_by = 'list_ID',
- feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
- point_border_col = 'gray'
- )
+spatFeatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'niche cluster',
+ expression_values = "prob",
+ group_by = 'list_ID',
+ feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
+ point_border_col = 'gray'
+ )
16.1.3.3 binarized niche cluster
-spatPlot2D(giotto_obj,
- spat_unit = "cell",
- group_by = 'list_ID',
- cell_color = "NicheCluster",
- color_as_factor = T,
- point_size = 1,
- point_border_stroke = NA,
- title="Niche cluster")
+spatPlot2D(giotto_obj,
+ spat_unit = "cell",
+ group_by = 'list_ID',
+ cell_color = "NicheCluster",
+ color_as_factor = T,
+ point_size = 1,
+ point_border_stroke = NA,
+ title="Niche cluster")
16.1.3.5 NT score
-spatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'rna',
- group_by = 'list_ID',
- cell_color = "NTScore",
- color_as_factor = FALSE,
- cell_color_gradient = "turbo",
- point_size = 1,
- point_border_stroke = NA
- )
+spatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ group_by = 'list_ID',
+ cell_color = "NTScore",
+ color_as_factor = FALSE,
+ cell_color_gradient = "turbo",
+ point_size = 1,
+ point_border_stroke = NA
+ )
-
+
diff --git a/interoperability-with-other-frameworks.html b/interoperability-with-other-frameworks.html
index 7050533..347d3af 100644
--- a/interoperability-with-other-frameworks.html
+++ b/interoperability-with-other-frameworks.html
@@ -521,10 +521,10 @@ 15 Interoperability with other fr
15.1 Load Giotto object
To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process:
-
+
Load a Giotto mini Visium object, which will be used for demonstrating interoperability.
-
+
15.2 Seurat
@@ -533,99 +533,99 @@ 15.2 Seurat
15.2.1 Conversion of Giotto Object to Seurat Object
To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function
-
+
Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data.
-gToS <- giottoToSeuratV5(gobject = gobject,
- spat_unit = "cell")
-
-plot1 <- VlnPlot(gToS,
- features = "nCount_rna",
- pt.size = 0.1) +
- NoLegend()
-
-plot2 <- SpatialFeaturePlot(gToS,
- features = "nCount_rna",
- pt.size.factor = 2) +
- theme(legend.position = "right")
-
-wrap_plots(plot1, plot2)
+gToS <- giottoToSeuratV5(gobject = gobject,
+ spat_unit = "cell")
+
+plot1 <- VlnPlot(gToS,
+ features = "nCount_rna",
+ pt.size = 0.1) +
+ NoLegend()
+
+plot2 <- SpatialFeaturePlot(gToS,
+ features = "nCount_rna",
+ pt.size.factor = 2) +
+ theme(legend.position = "right")
+
+wrap_plots(plot1, plot2)
15.2.1.1 Apply SCTransform
We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function.
-
+
15.2.1.2 Dimension Reduction
We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object:
-
+
15.2.2 Conversion of Seurat object Back to Giotto Object
To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks.
-giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
- spatial_assay = "rna",
- dim_reduction = c("pca", "umap"),
- sp_network = "Delaunay_network",
- nn_network = c("sNN.pca", "custom_NN" ))
+giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
+ spatial_assay = "rna",
+ dim_reduction = c("pca", "umap"),
+ sp_network = "Delaunay_network",
+ nn_network = c("sNN.pca", "custom_NN" ))
15.2.2.1 Clustering and Plotting UMAP
Here we perform K-means clustering on the UMAP results obtained from the Seurat object:
-## k-means clustering
-giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
- dim_reduction_to_use = "pca")
-
-#Plot UMAP post-clustering to visualize kmeans
-graph2 <- Giotto::plotUMAP(
- gobject = giottoFromSeurat,
- cell_color = "kmeans",
- show_NN_network = TRUE,
- point_size = 2.5
-)
+## k-means clustering
+giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
+ dim_reduction_to_use = "pca")
+
+#Plot UMAP post-clustering to visualize kmeans
+graph2 <- Giotto::plotUMAP(
+ gobject = giottoFromSeurat,
+ cell_color = "kmeans",
+ show_NN_network = TRUE,
+ point_size = 2.5
+)
15.2.2.2 Spatial CoExpression
We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function:
-ranktest <- binSpect(giottoFromSeurat,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "Delaunay_network")
-
-ext_spatial_genes <- ranktest[1:300,]$feats
-
-spat_cor_netw_DT <- detectSpatialCorFeats(
- giottoFromSeurat,
- method = "network",
- spatial_network_name = "Delaunay_network",
- subset_feats = ext_spatial_genes)
-
-top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
- feats = "Dsp",
- show_top_feats = 10)
-
-spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
- name = "spat_netw_clus",
- k = 7)
-heatmSpatialCorFeats(
- giottoFromSeurat,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL),
- save_plot = TRUE,
- show_plot = TRUE,
- return_plot = FALSE,
- save_param = list(base_height = 6, base_width = 8, units = 'cm'))
+ranktest <- binSpect(giottoFromSeurat,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "Delaunay_network")
+
+ext_spatial_genes <- ranktest[1:300,]$feats
+
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ giottoFromSeurat,
+ method = "network",
+ spatial_network_name = "Delaunay_network",
+ subset_feats = ext_spatial_genes)
+
+top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
+ feats = "Dsp",
+ show_top_feats = 10)
+
+spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
+ name = "spat_netw_clus",
+ k = 7)
+heatmSpatialCorFeats(
+ giottoFromSeurat,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ return_plot = FALSE,
+ save_param = list(base_height = 6, base_width = 8, units = 'cm'))
@@ -635,60 +635,60 @@ 15.3 SpatialExperiment
To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries.
-library(SpatialExperiment)
-library(ggspavis)
-library(pheatmap)
-library(scater)
-library(scran)
-library(nnSVG)
+library(SpatialExperiment)
+library(ggspavis)
+library(pheatmap)
+library(scater)
+library(scran)
+library(nnSVG)
15.3.1 Convert Giotto Object to SpatialExperiment Object
To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function.
-
+
The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use:
-
+
15.3.1.1 Identify top spatially variable genes with nnSVG
We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate:
-# One of the assays should be "logcounts"
-# We rename the normalized assay to "logcounts"
-assayNames(spe)[[2]] <- "logcounts"
-
-# Create model matrix for leiden clustering labels
-X <- model.matrix(~ colData(spe)$leiden_clus)
-dim(X)
+# One of the assays should be "logcounts"
+# We rename the normalized assay to "logcounts"
+assayNames(spe)[[2]] <- "logcounts"
+
+# Create model matrix for leiden clustering labels
+X <- model.matrix(~ colData(spe)$leiden_clus)
+dim(X)
- Run nnSVG
This step will take several minutes to run
-
+
15.3.2 Conversion of SpatialExperiment object back to Giotto
We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object.
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-print(giottoFromSPE)
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+print(giottoFromSPE)
15.3.2.1 Plotting top genes from nnSVG in Giotto
Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object.
-ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
- output = "data.table")
-
-ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
-spatFeatPlot2D(giottoFromSPE,
- expression_values = "scaled_rna_cell",
- feats = ext_spatial_genes[1:4],
- point_size = 2)
+ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
+ output = "data.table")
+
+ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
+spatFeatPlot2D(giottoFromSPE,
+ expression_values = "scaled_rna_cell",
+ feats = ext_spatial_genes[1:4],
+ point_size = 2)
@@ -701,97 +701,97 @@ 15.4 AnnData
15.4.1 Load Required Libraries
To start, we need to load the necessary libraries, including reticulate for interfacing with Python.
-
+
15.4.2 Specify Path for Results
First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions.
-# Specify path to which results may be saved
-results_directory <- "results/03_session4/giotto_anndata_conversion/"
-
-instrs <- showGiottoInstructions(gobject)
-
-mini_gobject <- replaceGiottoInstructions(gobject = gobject,
- instructions = instrs)
+# Specify path to which results may be saved
+results_directory <- "results/03_session4/giotto_anndata_conversion/"
+
+instrs <- showGiottoInstructions(gobject)
+
+mini_gobject <- replaceGiottoInstructions(gobject = gobject,
+ instructions = instrs)
15.4.2.1 Create Default kNN Network
We will create a k-nearest neighbor (kNN) network using mostly default parameters.
-
+
15.4.3 Giotto To AnnData
To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData()
-
+
Next, we import scanpy and perform a series of preprocessing steps on the AnnData object.
-scanpy <- import("scanpy")
-
-adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
-
-# Normalize total counts per cell
-scanpy$pp$normalize_total(adata, target_sum=1e4)
-
-# Log-transform the data
-scanpy$pp$log1p(adata)
-
-# Perform PCA
-scanpy$pp$pca(adata, n_comps=40L)
-
-# Compute the neighborhood graph
-scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
-
-# Run UMAP
-scanpy$tl$umap(adata)
-
-# Save the processed AnnData object
-adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
-
-processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
+scanpy <- import("scanpy")
+
+adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
+
+# Normalize total counts per cell
+scanpy$pp$normalize_total(adata, target_sum=1e4)
+
+# Log-transform the data
+scanpy$pp$log1p(adata)
+
+# Perform PCA
+scanpy$pp$pca(adata, n_comps=40L)
+
+# Compute the neighborhood graph
+scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
+
+# Run UMAP
+scanpy$tl$umap(adata)
+
+# Save the processed AnnData object
+adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
+
+processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
15.4.4 Convert AnnData to Giotto
Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto.
-
+
15.4.4.1 UMAP Visualization
Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object.
-Giotto::plotUMAP(
- gobject = giottoFromAnndata,
- dim_reduction_name = "umap.ad",
- cell_color = "leiden_clus",
- point_size = 3
-)
+Giotto::plotUMAP(
+ gobject = giottoFromAnndata,
+ dim_reduction_name = "umap.ad",
+ cell_color = "leiden_clus",
+ point_size = 3
+)
15.5 Create mini Vizgen object
-mini_gobject <- loadGiottoMini(dataset = "vizgen",
- python_path = NULL)
-
-mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
- instructions = instrs)
-mini_gobject <- createNearestNetwork(gobject = mini_gobject,
- spat_unit = "aggregate",
- feat_type = "rna",
- type = "kNN",
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap",
- k = 6,
- name = "new_network")
+mini_gobject <- loadGiottoMini(dataset = "vizgen",
+ python_path = NULL)
+
+mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
+ instructions = instrs)
+mini_gobject <- createNearestNetwork(gobject = mini_gobject,
+ spat_unit = "aggregate",
+ feat_type = "rna",
+ type = "kNN",
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap",
+ k = 6,
+ name = "new_network")
Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details.
-
+
diff --git a/introduction-to-the-giotto-package.html b/introduction-to-the-giotto-package.html
index e22d6d1..0ab2014 100644
--- a/introduction-to-the-giotto-package.html
+++ b/introduction-to-the-giotto-package.html
@@ -546,19 +546,58 @@ 4.3 Installation + python environ
4.3.1 Giotto installation
Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository.
Much of this already covered in Section 2.2, but the highlights are:
-
-- Ensure your system
-
-To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak.
+
+4.3.1.1 System prerequisites
+
+- for windows, Rtools needs to be installed
+- a major dependency terra needs
GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux
+
+
+
+4.3.1.2 Installation of released version
+To install the currently released version of Giotto in a single step:
+This should automatically install all the Giotto dependencies and other Giotto module packages.
+
+
+4.3.1.3 Installation of dev branch Giotto packages
+You can install dev branch versions by using devtools::install_github()
+Core module dev branchs:
+
+- “drieslab/Giotto@suite_dev”
+- “drieslab/GiottoVisuals@dev”
+- “drieslab/GiottoClass@dev”
+- “drieslab/GiottoUtils@dev”
+
+
+pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages.
+
+
+4.3.1.4 Common install issues
+If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work.
+
+If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it.
+
+## # Downloading packages -------------------------------------------------------
+## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s]
+## Successfully downloaded 1 package in 1.2 seconds.
+##
+## The following package(s) will be installed:
+## - irlba [2.3.5.1]
+## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu".
+##
+## # Installing packages --------------------------------------------------------
+## - Installing irlba ... OK [built from source and cached in 5.7s]
+## Successfully installed 1 package in 5.7 seconds.
+
4.3.2 Python environment
4.3.2.1 Default installation
In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment.
-
+
Two things are needed for python to work:
- A conda (e.g. miniconda or anaconda) installation which is the package and environment management system.
@@ -580,17 +619,17 @@ 4.3.2.1 Default installation
4.3.2.2 Custom installs
Custom python environments can be made by first setting up a new environment and establishing the name and python version to use.
-
+
Following that, one or more python packages to install can be added to the environment.
-reticulate::py_install(
- pip = TRUE,
- envname = '[name of env]',
- packages = c(
- "package1",
- "package2",
- "..."
- )
-)
+reticulate::py_install(
+ pip = TRUE,
+ envname = '[name of env]',
+ packages = c(
+ "package1",
+ "package2",
+ "..."
+ )
+)
Once an environment has been set up, Giotto can hook into it.
@@ -612,17 +651,56 @@ 4.3.2.3 Using a specific environm
Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto.
You would run file.edit("~/.Rprofile") and then add options("giotto.py_path" = "[path to env or envname]") as a line so that it is automatically set at the start of each session.
If a specific environment should only be used a couple times then method 1 is easiest:
-
+
To check which conda environments exist on your machine:
-
+
Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running:
-
+
4.4 Giotto instructions
-Giotto
+Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are:
+
+- python_path - when set, will activate a python environment
+- save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render.
+- save_plot - whether to save plots to the
save_dir
+- return_plot - whether to return the plot objects. When FALSE, only NULL is returned
+- show_plot - whether to show the plot in the viewer
+
+These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function.
+library(Giotto)
+save_dir <- "results/01_session2/"
+
+# this call will also intialize the python env
+instrs <- createGiottoInstructions(
+ save_dir = save_dir, # working directory is the default
+ show_plot = FALSE,
+ save_plot = TRUE,
+ return_plot = FALSE,
+ python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default
+)
+force(instrs)
+Giotto object creation functions all have an instructions param for passing in instructions objects.
+giotto objects will also respond to the instructions() generic.
+test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object
+
+# example plot
+g <- GiottoData::loadGiottoMini("visium")
+instructions(g) <- instrs
+instructions(g, "show_plot") # instructions say not to plot to viewer
+spatPlot2D(g, show_image = TRUE, image_name = "image")
+# instead it will directly write to the results folder
+As an example, you can also set individual instructions
+
+
+
+
+Figure 4.2: example image output
+
+
diff --git a/multi-omics-integration.html b/multi-omics-integration.html
index 47cad23..e76e25a 100644
--- a/multi-omics-integration.html
+++ b/multi-omics-integration.html
@@ -520,14 +520,14 @@ 14 Multi-omics integration
14.1 The CytAssist technology
The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image.
-
+
Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics.
The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument.
-
+
Figure 14.2: CytAssist workflow. Source: 10X genomics.
@@ -542,19 +542,19 @@
14.2 Introduction to the spatial
14.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/03_session3")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
+dir.create("data/03_session3")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”.
-
+
14.4 Create the Giotto object
@@ -564,124 +564,124 @@ 14.4 Create the Giotto objectx,y(,z) coordinates for cells or spots (or the path to)
createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/03_session3/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-# Provide the path to the data folder
-data_path <- "data/03_session3/"
-
-# Create object directly from the data folder
-visium_tonsil <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/03_session3/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+# Provide the path to the data folder
+data_path <- "data/03_session3/"
+
+# Create object directly from the data folder
+visium_tonsil <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
- Print the information of the object, note that both rna and protein are listed in the expression slot.
-
+
14.5 Subset on spots that were covered by tissue
-spatPlot2D(
- gobject = visium_tonsil,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE,
- image_name = "image"
-)
-
+spatPlot2D(
+ gobject = visium_tonsil,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE,
+ image_name = "image"
+)
+
Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object.
-
+
14.6 RNA processing
- Run the Filtering, normalization, and statistics steps using only the RNA feature.
-visium_tonsil <- filterGiotto(
- gobject = visium_tonsil,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil)
+visium_tonsil <- filterGiotto(
+ gobject = visium_tonsil,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1000,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil)
- Dimension reduction
Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs.
-visium_tonsil <- calculateHVF(gobject = visium_tonsil)
-
-visium_tonsil <- runPCA(gobject = visium_tonsil)
+visium_tonsil <- calculateHVF(gobject = visium_tonsil)
+
+visium_tonsil <- runPCA(gobject = visium_tonsil)
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the RNA-based Leiden clusters.
-
+
- Visualization
Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot.
-plotUMAP(gobject = visium_tonsil,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters.
Plot the spatial distribution of the RNA-based Leiden cluster per spot.
-spatPlot2D(gobject = visium_tonsil,
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.5: Spatial distribution of RNA-based Leiden clusters.
@@ -693,80 +693,80 @@
14.7 Protein processingvisium_tonsil <- filterGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein")
+visium_tonsil <- filterGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein")
- Dimension reduction
Calculate the principal components using all the proteins available in the dataset.
-
+
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the Protein-based Leiden clusters.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- resolution = 1,
- n_iterations = 1000)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ resolution = 1,
+ n_iterations = 1000)
- Visualization
Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters.
Plot the spatial distribution of the Protein-based Leiden clusters.
-spatPlot2D(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.7: Spatial distribution of Protein-based Leiden clusters.
@@ -779,68 +779,68 @@
14.8 Multi-omics integrationCalculate kNN
Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type.
-## RNA modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
-
-## Protein modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
+## RNA modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
+
+## Protein modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
- Run WNN
Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object.
-visium_tonsil <- runWNN(visium_tonsil,
- spat_unit = "cell",
- modality_1 = "rna",
- modality_2 = "protein",
- pca_name_modality_1 = "pca",
- pca_name_modality_2 = "protein.pca",
- k = 20,
- integrated_feat_type = NULL,
- matrix_result_name = NULL,
- w_name_modality_1 = NULL,
- w_name_modality_2 = NULL,
- verbose = TRUE)
+visium_tonsil <- runWNN(visium_tonsil,
+ spat_unit = "cell",
+ modality_1 = "rna",
+ modality_2 = "protein",
+ pca_name_modality_1 = "pca",
+ pca_name_modality_2 = "protein.pca",
+ k = 20,
+ integrated_feat_type = NULL,
+ matrix_result_name = NULL,
+ w_name_modality_1 = NULL,
+ w_name_modality_2 = NULL,
+ verbose = TRUE)
- Run Integrated umap
Calculate the UMAP using the weights of each feature per spot.
-visium_tonsil <- runIntegratedUMAP(visium_tonsil,
- modality1 = "rna",
- modality2 = "protein",
- spread = 7,
- min_dist = 1,
- force = FALSE)
+visium_tonsil <- runIntegratedUMAP(visium_tonsil,
+ modality1 = "rna",
+ modality2 = "protein",
+ spread = 7,
+ min_dist = 1,
+ force = FALSE)
- Calculate integrated clusters
Calculate the multiomics-based Leiden clusters using the weights of each feature per spot.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- nn_network_to_use = "kNN",
- network_name = "integrated_kNN",
- name = "integrated_leiden_clus",
- resolution = 1)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ nn_network_to_use = "kNN",
+ network_name = "integrated_kNN",
+ name = "integrated_leiden_clus",
+ resolution = 1)
- Visualize the integrated UMAP
Plot the integrated UMAP and color the spots using the integrated Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- dim_reduction_name = "integrated.umap",
- point_size = 2,
- title = "Integrated UMAP using Integrated Leiden clusters")
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ dim_reduction_name = "integrated.umap",
+ point_size = 2,
+ title = "Integrated UMAP using Integrated Leiden clusters")
+
Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters.
@@ -850,14 +850,14 @@
14.8 Multi-omics integrationVisualize spatial plot with integrated clusters
Plot the spatial distribution of the integrated Leiden clusters.
-spatPlot2D(visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- point_size = 3,
- show_image = TRUE,
- title = "Integrated Leiden clustering")
-
+spatPlot2D(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ point_size = 3,
+ show_image = TRUE,
+ title = "Integrated Leiden clustering")
+
Figure 14.9: Spatial distribution of the integrated Leiden clusters.
@@ -866,85 +866,85 @@
14.8 Multi-omics integration
14.9 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
- [7] matrixStats_1.3.0 compiler_4.4.1
- [9] png_0.1-8 systemfonts_1.1.0
- [11] vctrs_0.6.5 reshape2_1.4.4
- [13] stringr_1.5.1 pkgconfig_2.0.3
- [15] SpatialExperiment_1.14.0 crayon_1.5.3
- [17] fastmap_1.2.0 backports_1.5.0
- [19] magick_2.8.4 XVector_0.44.0
- [21] labeling_0.4.3 utf8_1.2.4
- [23] rmarkdown_2.27 UCSC.utils_1.0.0
- [25] ragg_1.3.2 purrr_1.0.2
- [27] xfun_0.46 beachmat_2.20.0
- [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
- [31] jsonlite_1.8.8 DelayedArray_0.30.1
- [33] BiocParallel_1.38.0 terra_1.7-78
- [35] irlba_2.3.5.1 parallel_4.4.1
- [37] R6_2.5.1 stringi_1.8.4
- [39] RColorBrewer_1.1-3 reticulate_1.38.0
- [41] parallelly_1.37.1 GenomicRanges_1.56.1
- [43] scattermore_1.2 Rcpp_1.0.13
- [45] bookdown_0.40 SummarizedExperiment_1.34.0
- [47] knitr_1.48 future.apply_1.11.2
- [49] R.utils_2.12.3 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
+ [7] matrixStats_1.3.0 compiler_4.4.1
+ [9] png_0.1-8 systemfonts_1.1.0
+ [11] vctrs_0.6.5 reshape2_1.4.4
+ [13] stringr_1.5.1 pkgconfig_2.0.3
+ [15] SpatialExperiment_1.14.0 crayon_1.5.3
+ [17] fastmap_1.2.0 backports_1.5.0
+ [19] magick_2.8.4 XVector_0.44.0
+ [21] labeling_0.4.3 utf8_1.2.4
+ [23] rmarkdown_2.27 UCSC.utils_1.0.0
+ [25] ragg_1.3.2 purrr_1.0.2
+ [27] xfun_0.46 beachmat_2.20.0
+ [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
+ [31] jsonlite_1.8.8 DelayedArray_0.30.1
+ [33] BiocParallel_1.38.0 terra_1.7-78
+ [35] irlba_2.3.5.1 parallel_4.4.1
+ [37] R6_2.5.1 stringi_1.8.4
+ [39] RColorBrewer_1.1-3 reticulate_1.38.0
+ [41] parallelly_1.37.1 GenomicRanges_1.56.1
+ [43] scattermore_1.2 Rcpp_1.0.13
+ [45] bookdown_0.40 SummarizedExperiment_1.34.0
+ [47] knitr_1.48 future.apply_1.11.2
+ [49] R.utils_2.12.3 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/reference-keys.txt b/reference-keys.txt
index a17604d..555f956 100644
--- a/reference-keys.txt
+++ b/reference-keys.txt
@@ -4,115 +4,116 @@ fig:unnamed-chunk-14
fig:unnamed-chunk-15
fig:unnamed-chunk-16
fig:unnamed-chunk-18
-fig:unnamed-chunk-30
-fig:unnamed-chunk-35
-fig:unnamed-chunk-38
-fig:unnamed-chunk-40
+fig:unnamed-chunk-32
+fig:unnamed-chunk-37
fig:unnamed-chunk-42
-fig:unnamed-chunk-44
+fig:unnamed-chunk-45
+fig:unnamed-chunk-47
fig:unnamed-chunk-49
fig:unnamed-chunk-51
-fig:unnamed-chunk-53
-fig:unnamed-chunk-55
-fig:unnamed-chunk-59
-fig:unnamed-chunk-61
-fig:unnamed-chunk-63
+fig:unnamed-chunk-56
+fig:unnamed-chunk-58
+fig:unnamed-chunk-60
+fig:unnamed-chunk-62
fig:unnamed-chunk-66
-fig:unnamed-chunk-69
-fig:unnamed-chunk-74
+fig:unnamed-chunk-68
+fig:unnamed-chunk-70
+fig:unnamed-chunk-73
fig:unnamed-chunk-76
-fig:unnamed-chunk-78
-fig:unnamed-chunk-80
-fig:unnamed-chunk-82
-fig:unnamed-chunk-84
+fig:unnamed-chunk-81
+fig:unnamed-chunk-83
+fig:unnamed-chunk-85
fig:unnamed-chunk-87
+fig:unnamed-chunk-89
+fig:unnamed-chunk-91
fig:unnamed-chunk-94
-fig:unnamed-chunk-96
-fig:unnamed-chunk-98
fig:unnamed-chunk-101
fig:unnamed-chunk-103
fig:unnamed-chunk-105
+fig:unnamed-chunk-108
+fig:unnamed-chunk-110
fig:unnamed-chunk-112
-fig:unnamed-chunk-114
-fig:unnamed-chunk-118
-fig:unnamed-chunk-120
-fig:unnamed-chunk-123
-fig:unnamed-chunk-128
-fig:unnamed-chunk-131
-fig:unnamed-chunk-134
-fig:unnamed-chunk-137
+fig:unnamed-chunk-119
+fig:unnamed-chunk-121
+fig:unnamed-chunk-125
+fig:unnamed-chunk-127
+fig:unnamed-chunk-130
+fig:unnamed-chunk-135
+fig:unnamed-chunk-138
fig:unnamed-chunk-141
-fig:unnamed-chunk-143
-fig:unnamed-chunk-147
+fig:unnamed-chunk-144
+fig:unnamed-chunk-148
fig:unnamed-chunk-150
-fig:unnamed-chunk-152
+fig:unnamed-chunk-154
+fig:unnamed-chunk-157
fig:unnamed-chunk-159
-fig:unnamed-chunk-161
-fig:unnamed-chunk-163
-fig:unnamed-chunk-165
-fig:unnamed-chunk-167
-fig:unnamed-chunk-169
-fig:unnamed-chunk-171
+fig:unnamed-chunk-166
+fig:unnamed-chunk-168
+fig:unnamed-chunk-170
+fig:unnamed-chunk-172
fig:unnamed-chunk-174
-fig:unnamed-chunk-175
fig:unnamed-chunk-176
-fig:unnamed-chunk-185
-fig:unnamed-chunk-191
-fig:unnamed-chunk-194
+fig:unnamed-chunk-178
+fig:unnamed-chunk-181
+fig:unnamed-chunk-182
+fig:unnamed-chunk-183
+fig:unnamed-chunk-192
+fig:unnamed-chunk-198
fig:unnamed-chunk-201
-fig:unnamed-chunk-204
-fig:unnamed-chunk-206
-fig:unnamed-chunk-210
-fig:unnamed-chunk-212
-fig:unnamed-chunk-215
+fig:unnamed-chunk-208
+fig:unnamed-chunk-211
+fig:unnamed-chunk-213
fig:unnamed-chunk-217
fig:unnamed-chunk-219
-fig:unnamed-chunk-221
+fig:unnamed-chunk-222
fig:unnamed-chunk-224
-fig:unnamed-chunk-225
-fig:unnamed-chunk-227
-fig:unnamed-chunk-229
-fig:unnamed-chunk-230
+fig:unnamed-chunk-226
+fig:unnamed-chunk-228
+fig:unnamed-chunk-231
fig:unnamed-chunk-232
fig:unnamed-chunk-234
fig:unnamed-chunk-236
-fig:unnamed-chunk-275
-fig:unnamed-chunk-279
-fig:unnamed-chunk-281
-fig:unnamed-chunk-283
+fig:unnamed-chunk-237
+fig:unnamed-chunk-239
+fig:unnamed-chunk-241
+fig:unnamed-chunk-243
+fig:unnamed-chunk-282
fig:unnamed-chunk-286
fig:unnamed-chunk-288
fig:unnamed-chunk-290
-fig:unnamed-chunk-292
-fig:unnamed-chunk-294
+fig:unnamed-chunk-293
+fig:unnamed-chunk-295
fig:unnamed-chunk-297
fig:unnamed-chunk-299
fig:unnamed-chunk-301
-fig:unnamed-chunk-303
-fig:unnamed-chunk-305
-fig:unnamed-chunk-365
-fig:unnamed-chunk-366
+fig:unnamed-chunk-304
+fig:unnamed-chunk-306
+fig:unnamed-chunk-308
+fig:unnamed-chunk-310
+fig:unnamed-chunk-312
fig:unnamed-chunk-372
+fig:unnamed-chunk-373
fig:unnamed-chunk-379
-fig:unnamed-chunk-381
-fig:unnamed-chunk-387
-fig:unnamed-chunk-389
-fig:unnamed-chunk-395
-fig:unnamed-chunk-397
-fig:unnamed-chunk-462
-fig:unnamed-chunk-465
+fig:unnamed-chunk-386
+fig:unnamed-chunk-388
+fig:unnamed-chunk-394
+fig:unnamed-chunk-396
+fig:unnamed-chunk-402
+fig:unnamed-chunk-404
fig:unnamed-chunk-469
-fig:unnamed-chunk-470
fig:unnamed-chunk-472
-fig:unnamed-chunk-474
fig:unnamed-chunk-476
-fig:unnamed-chunk-478
+fig:unnamed-chunk-477
fig:unnamed-chunk-479
fig:unnamed-chunk-481
+fig:unnamed-chunk-483
fig:unnamed-chunk-485
+fig:unnamed-chunk-486
fig:unnamed-chunk-488
-fig:unnamed-chunk-490
fig:unnamed-chunk-492
+fig:unnamed-chunk-495
+fig:unnamed-chunk-497
+fig:unnamed-chunk-499
giotto-suite-workshop-2024
instructors
topics-and-schedule
@@ -153,6 +154,10 @@ presentation-1
ecosystem
installation-python-environment
giotto-installation-1
+system-prerequisites
+installation-of-released-version
+installation-of-dev-branch-giotto-packages
+common-install-issues
python-environment
default-installation
custom-installs
diff --git a/search_index.json b/search_index.json
index 2736898..72ab7c3 100644
--- a/search_index.json
+++ b/search_index.json
@@ -1 +1 @@
-[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: Ensure your system To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak. pak::pak("drieslab/Giotto") 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
+[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: 4.3.1.1 System prerequisites for windows, Rtools needs to be installed a major dependency terra needs GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux 4.3.1.2 Installation of released version To install the currently released version of Giotto in a single step: pak::pak("drieslab/Giotto") This should automatically install all the Giotto dependencies and other Giotto module packages. 4.3.1.3 Installation of dev branch Giotto packages You can install dev branch versions by using devtools::install_github() Core module dev branchs: “drieslab/Giotto@suite_dev” “drieslab/GiottoVisuals@dev” “drieslab/GiottoClass@dev” “drieslab/GiottoUtils@dev” devtools::install_github("drieslab/GiottoClass@dev") pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages. 4.3.1.4 Common install issues If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work. devtools::install_version("Matrix", version = "1.6-5") If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it. install.packages("irlba", type = "source") ## # Downloading packages ------------------------------------------------------- ## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s] ## Successfully downloaded 1 package in 1.2 seconds. ## ## The following package(s) will be installed: ## - irlba [2.3.5.1] ## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu". ## ## # Installing packages -------------------------------------------------------- ## - Installing irlba ... OK [built from source and cached in 5.7s] ## Successfully installed 1 package in 5.7 seconds. 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are: python_path - when set, will activate a python environment save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render. save_plot - whether to save plots to the save_dir return_plot - whether to return the plot objects. When FALSE, only NULL is returned show_plot - whether to show the plot in the viewer These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function. library(Giotto) save_dir <- "results/01_session2/" # this call will also intialize the python env instrs <- createGiottoInstructions( save_dir = save_dir, # working directory is the default show_plot = FALSE, save_plot = TRUE, return_plot = FALSE, python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default ) force(instrs) Giotto object creation functions all have an instructions param for passing in instructions objects. giotto objects will also respond to the instructions() generic. test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object # example plot g <- GiottoData::loadGiottoMini("visium") instructions(g) <- instrs instructions(g, "show_plot") # instructions say not to plot to viewer spatPlot2D(g, show_image = TRUE, image_name = "image") # instead it will directly write to the results folder As an example, you can also set individual instructions instructions(g, "show_plot") <- TRUE spatPlot2D(g, show_image = TRUE, image_name = "image") Figure 4.2: example image output "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
diff --git a/spatial-multi-modal-analysis.html b/spatial-multi-modal-analysis.html
index 5bdab74..c78f952 100644
--- a/spatial-multi-modal-analysis.html
+++ b/spatial-multi-modal-analysis.html
@@ -563,48 +563,48 @@ 13.2.3 Spatial classes:# load in data
-library(Giotto)
-
-g <- GiottoData::loadGiottoMini("vizgen")
-
-activeSpatUnit(g) <- "aggregate"
-
-gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE)
-
-gimg <- getGiottoImage(g)
+
13.3 Examples of the simple transforms with a giottoPolygon
-rain <- rainbow(nrow(gpoly))
-
-line_width <- 0.3
-
-# par to setup the grid plotting layout
-p <- par(no.readonly = TRUE)
-par(mfrow=c(3,3))
-gpoly |>
- plot(main = "no transform", col = rain, lwd = line_width)
-
-gpoly |> spatShift(dx = 1000) |>
- plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
-
-gpoly |> spin(45) |>
- plot(main = "spin(45)", col = rain, lwd = line_width)
-
-gpoly |> rescale(fx = 10, fy = 6) |>
- plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
-
-gpoly |> flip(direction = "vertical") |>
- plot(main = "flip()", col = rain, lwd = line_width)
-
-gpoly |> t() |>
- plot(main = "t()", col = rain, lwd = line_width)
-
-gpoly |> shear(fx = 0.5) |>
- plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
-par(p)
+rain <- rainbow(nrow(gpoly))
+
+line_width <- 0.3
+
+# par to setup the grid plotting layout
+p <- par(no.readonly = TRUE)
+par(mfrow=c(3,3))
+gpoly |>
+ plot(main = "no transform", col = rain, lwd = line_width)
+
+gpoly |> spatShift(dx = 1000) |>
+ plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
+
+gpoly |> spin(45) |>
+ plot(main = "spin(45)", col = rain, lwd = line_width)
+
+gpoly |> rescale(fx = 10, fy = 6) |>
+ plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
+
+gpoly |> flip(direction = "vertical") |>
+ plot(main = "flip()", col = rain, lwd = line_width)
+
+gpoly |> t() |>
+ plot(main = "t()", col = rain, lwd = line_width)
+
+gpoly |> shear(fx = 0.5) |>
+ plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
+par(p)
@@ -617,20 +617,20 @@ 13.4 Affine transforms# create affine2d
-aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
+# create affine2d
+aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set.
-
-# append several simple transforms
-aff <- aff |>
- spatShift(dx = 1000) |>
- spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- flip(direction = "vertical") |>
- t() |>
- shear(fx = 0.5)
-force(aff)
+
+# append several simple transforms
+aff <- aff |>
+ spatShift(dx = 1000) |>
+ spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ flip(direction = "vertical") |>
+ t() |>
+ shear(fx = 0.5)
+force(aff)
<affine2d>
anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax)
rotate : -0.785398163397448 (rad)
@@ -639,35 +639,35 @@ 13.4 Affine transformsplot(aff)
+
We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests.
-
+
13.5 Image transforms
Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting.
spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image.
-# affine transform of image info matches with polygon info
-gimg |> affine(aff) |> plot()
-gpoly |> affine(aff) |>
- plot(add = TRUE,
- border = "cyan",
- lwd = 0.3)
+# affine transform of image info matches with polygon info
+gimg |> affine(aff) |> plot()
+gpoly |> affine(aff) |>
+ plot(add = TRUE,
+ border = "cyan",
+ lwd = 0.3)
-# affine of the giotto object
-g |> affine(aff) |>
- spatInSituPlotPoints(
- show_image = TRUE,
- feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
- feats_color_code = rainbow(4),
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- point_size = 0.1,
- use_overlap = FALSE
- )
+# affine of the giotto object
+g |> affine(aff) |>
+ spatInSituPlotPoints(
+ show_image = TRUE,
+ feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
+ feats_color_code = rainbow(4),
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ point_size = 0.1,
+ use_overlap = FALSE
+ )
Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used.
@@ -687,256 +687,256 @@ 13.6.1 Example dataset: Xenium Br
– cooresponding centroid locations
The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance.
-library(Giotto)
-instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
- save_plot = TRUE,
- show_plot = TRUE)
-options(timeout = 999999)
-download_dir <-file.path(getwd(),'/data/03_session2/')
-destfile <- file.path(download_dir,'Multimodal_registration.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
-Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
-Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
+library(Giotto)
+instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
+ save_plot = TRUE,
+ show_plot = TRUE)
+options(timeout = 999999)
+download_dir <-file.path(getwd(),'/data/03_session2/')
+destfile <- file.path(download_dir,'Multimodal_registration.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
+Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
+Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
13.6.2 Target Coordinate system
Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system.
-xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
-xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
-xen_cell_pl
+xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
+xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+xen_cell_pl
13.6.3 Visium to register
Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration
-G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
- gene_column_index = 2,
- png_name = 'tissue_hires_image.png',
- instructions = NULL)
-# In the meantime, calculate statistics for easier plot showing
-G_visium <- normalizeGiotto(G_visium)
-G_visium <- addStatistics(G_visium)
-V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
-V_origin
+G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
+ gene_column_index = 2,
+ png_name = 'tissue_hires_image.png',
+ instructions = NULL)
+# In the meantime, calculate statistics for easier plot showing
+G_visium <- normalizeGiotto(G_visium)
+G_visium <- addStatistics(G_visium)
+V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
+V_origin
The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform.
-# create affine2d
-aff <- affine(diag(c(1,1)))
-aff <- aff |>
- spin(90) |>
- flip(direction = "horizontal")
-force(aff)
-
-# Apply the transform
-V_tansformed <- affine(G_visium,aff)
-spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
-spatplot_to_register
+# create affine2d
+aff <- affine(diag(c(1,1)))
+aff <- aff |>
+ spin(90) |>
+ flip(direction = "horizontal")
+force(aff)
+
+# Apply the transform
+V_tansformed <- affine(G_visium,aff)
+spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
+spatplot_to_register
Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images.
Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app.
-
+
Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result.
-landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
-affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
-
-V_final <- affine(G_visium,affine_mtx %*% aff@affine)
-
-spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
-spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
+affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
+
+V_final <- affine(G_visium,affine_mtx %*% aff@affine)
+
+spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
+spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.3.1 Create Pseudo Visium dataset for comparison
Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance.
-V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
-spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
-spatPlot2D(V_final,
- show_network = T,
- network_color = 'blue',
- point_size = 1)
-center_to_center <- min(spat_network$distance)
-radius <- center_to_center*55/200
+V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
+spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
+spatPlot2D(V_final,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1)
+center_to_center <- min(spat_network$distance)
+radius <- center_to_center*55/200
Now we get the Pseudo Visium polygons
-Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
-stamp_dt <- circleVertices(radius = radius, npoints = 100)
-pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
-pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
-plot(pseudo_visium_poly)
+Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
+stamp_dt <- circleVertices(radius = radius, npoints = 100)
+pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
+pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
+plot(pseudo_visium_poly)
Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points
-xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
-gpoints <- createGiottoPoints(xen_transcripts)
-Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
- gpolygons = list('visium' = pseudo_visium_poly))
+xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
+gpoints <- createGiottoPoints(xen_transcripts)
+Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
+ gpolygons = list('visium' = pseudo_visium_poly))
Get gene expression information by overlapping polygon to points
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'visium')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "visium",
- feat_info = "rna",
- aggr_function = "sum")
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'visium')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "visium",
+ feat_info = "rna",
+ aggr_function = "sum")
Manipulate the expression for plotting
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'visium',
- name = 'plot')
-
-spatFeatPlot2D(Xen_obj,
- point_size = 3.5,
- expression_values = 'plot',
- show_image = F,
- feats = 'ERBB2')
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ name = 'plot')
+
+spatFeatPlot2D(Xen_obj,
+ point_size = 3.5,
+ expression_values = 'plot',
+ show_image = F,
+ feats = 'ERBB2')
Subset the registered Visium and plot same gene
-#get the extent of giotto points, xmin, xmax, ymin, ymax
-subset_extent <- ext(gpoints@spatVector)
-sub_visium <- subsetGiottoLocs(V_final,
- x_min = subset_extent[1],
- x_max = subset_extent[2],
- y_min = subset_extent[3],
- y_max = subset_extent[4])
-spatFeatPlot2D(sub_visium,
- point_size = 2,
- expression_values = 'scaled',
- show_image = F,
- feats = 'ERBB2')
+#get the extent of giotto points, xmin, xmax, ymin, ymax
+subset_extent <- ext(gpoints@spatVector)
+sub_visium <- subsetGiottoLocs(V_final,
+ x_min = subset_extent[1],
+ x_max = subset_extent[2],
+ y_min = subset_extent[3],
+ y_max = subset_extent[4])
+spatFeatPlot2D(sub_visium,
+ point_size = 2,
+ expression_values = 'scaled',
+ show_image = F,
+ feats = 'ERBB2')
13.6.4 Register post-Xenium H&E and IF image
For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics.
-HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
- imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
- micron = 0.2125)
-plot(HE_xen)
+HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
+ imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
+ micron = 0.2125)
+plot(HE_xen)
The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use.
-HE_xen <- HE_xen |> flip(direction = "vertical")
-gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
-plot(gimg_rast)
-#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
+HE_xen <- HE_xen |> flip(direction = "vertical")
+gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
+plot(gimg_rast)
+#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots
-gg <- ggplot2::ggplot()
-pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
-pl + ggplot2::geom_smooth() +
- ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+gg <- ggplot2::ggplot()
+pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
+pl + ggplot2::geom_smooth() +
+ ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.4.1 Add registered image information and compare RNA vs protein expression
With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF
-CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
-HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
-Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
- largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
+CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
+HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
+Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
+ largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
Get the cell polygons, as Xenium and IF are both subcellular resolution
-cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
-colnames(cellpoly_dt) <- c('poly_ID','x','y')
-cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
-Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
+cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
+colnames(cellpoly_dt) <- c('poly_ID','x','y')
+cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
+Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
Compute the gene expression matrix by overlay the cell polygons and giotto points.
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'cell')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "cell",
- feat_info = "rna",
- aggr_function = "sum")
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'rna',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'ERBB2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'cell')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "cell",
+ feat_info = "rna",
+ aggr_function = "sum")
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'rna',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'ERBB2',
+ point_size = 0.05)
Now we overlay the HER2 expression from the raster image with the cell polygons.
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'cell',
- image_names = c('HER2','CD20'))
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'HER2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'cell',
+ image_names = c('HER2','CD20'))
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'HER2',
+ point_size = 0.05)
We can also overlay the protein expression to Visium spots
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'visium',
- image_names = c('HER2','CD20'))
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "visium",
- feat_info = "protein",
- aggr_function = "sum")
-
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'visium',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'visium',
- feats = 'HER2',
- point_size = 2)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'visium',
+ image_names = c('HER2','CD20'))
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "visium",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'visium',
+ feats = 'HER2',
+ point_size = 2)
@@ -944,29 +944,29 @@ 13.6.4.1 Add registered image inf
13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation
Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially.
Here, we provide an example of two compressed images to show the automatic alignment pipeline.
-
+
-IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
-terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
+IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
+terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform.
-estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
- flip_horizontal = T,
- use_single_channel = T,
- single_channel_number = 3),
- y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
- invert = T),
- plot_match = T,
- max_ratio = 0.5,estimate_fun = 'Projective')
+estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
+ flip_horizontal = T,
+ use_single_channel = T,
+ single_channel_number = 3),
+ y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
+ invert = T),
+ plot_match = T,
+ max_ratio = 0.5,estimate_fun = 'Projective')
Use the estimation, we can quickly visualize the transformation
-mtx <- as.matrix(estimation$params)
-transformed <- affine(IF, mtx)
-To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
-plot(HE)
-plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
-SIFT_registration_overlay
+mtx <- as.matrix(estimation$params)
+transformed <- affine(IF, mtx)
+To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
+plot(HE)
+plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
+SIFT_registration_overlay
diff --git a/spatial-proteomics-multiplexed-immunofluorescence.html b/spatial-proteomics-multiplexed-immunofluorescence.html
index 684d4e8..f550962 100644
--- a/spatial-proteomics-multiplexed-immunofluorescence.html
+++ b/spatial-proteomics-multiplexed-immunofluorescence.html
@@ -518,33 +518,33 @@ 11 Spatial proteomics: Multiplexe
Junxiang Xu
August 6th 2024
Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so.
-#Install the Giotto Environment with Cellpose, note that we only need to do it once
-reticulate::conda_create(envname = "giotto_cellpose",
- python_version = 3.8)
-#.re.restartR()
-reticulate::use_condaenv('giotto_cellpose')
-reticulate::py_install(
- pip = TRUE,
- envname = 'giotto_cellpose',
- packages = c(
- "pandas",
- "networkx",
- "python-igraph",
- "leidenalg",
- "scikit-learn",
- "cellpose",
- "smfishhmrf",
- 'tifffile',
- 'scikit-image'
- )
-)
-#.rs.restartR()
+#Install the Giotto Environment with Cellpose, note that we only need to do it once
+reticulate::conda_create(envname = "giotto_cellpose",
+ python_version = 3.8)
+#.re.restartR()
+reticulate::use_condaenv('giotto_cellpose')
+reticulate::py_install(
+ pip = TRUE,
+ envname = 'giotto_cellpose',
+ packages = c(
+ "pandas",
+ "networkx",
+ "python-igraph",
+ "leidenalg",
+ "scikit-learn",
+ "cellpose",
+ "smfishhmrf",
+ 'tifffile',
+ 'scikit-image'
+ )
+)
+#.rs.restartR()
Now, activate the Giotto python environment.
-#.rs.restartR()
-# Activate the Giotto python environment of your choice
-GiottoClass::set_giotto_python_path('giotto_cellpose')
-# Check if cellpose was successfully installed
-GiottoUtils::package_check('cellpose',repository = 'pip')
+#.rs.restartR()
+# Activate the Giotto python environment of your choice
+GiottoClass::set_giotto_python_path('giotto_cellpose')
+# Check if cellpose was successfully installed
+GiottoUtils::package_check('cellpose',repository = 'pip')
11.1 Spatial Proteomics Technologies
This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix.
@@ -557,58 +557,58 @@
11.2 Raw data type coming out of
11.2.1 Use ome.tiff as an example output data to begin with
OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore.
-library(Giotto)
-instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
- save_plot = TRUE,
- show_plot = TRUE,
- python_path = 'giotto_cellpose')
-options(timeout = 9999999)
-download_dir =file.path(getwd(),'/data/02_session4/')
-destfile = file.path(download_dir,'Lunaphore.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
-list.files(paste0(download_dir,'/Lunaphore'))
+library(Giotto)
+instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ python_path = 'giotto_cellpose')
+options(timeout = 9999999)
+download_dir =file.path(getwd(),'/data/02_session4/')
+destfile = file.path(download_dir,'Lunaphore.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
+list.files(paste0(download_dir,'/Lunaphore'))
We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML.
-img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
-img_meta
+img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
+img_meta
However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it.
-##Get channel information
-Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
-node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
-channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
-channel_df
+##Get channel information
+Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
+node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
+channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
+channel_df
11.2.2 Use single channel images as an example output data to begin with
Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like.
-#Create multichannel raster and extract each single channels
-Luna_terra = terra::rast(Luna)
-names(Luna_terra) <- channel_df$Name
-gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
-plot(gimg_DAPI)
+#Create multichannel raster and extract each single channels
+Luna_terra = terra::rast(Luna)
+names(Luna_terra) <- channel_df$Name
+gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
+plot(gimg_DAPI)
Extract and save the raster image for future use.
-single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
-if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
-for (i in 1:nrow(channel_df)){
- single_channel <- terra::subset(Luna_terra, i)
- terra::writeRaster(single_channel,
- filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
- overwrite=T)
-}
+single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
+if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
+for (i in 1:nrow(channel_df)){
+ single_channel <- terra::subset(Luna_terra, i)
+ terra::writeRaster(single_channel,
+ filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
+ overwrite=T)
+}
Create a list of GiottoLargeImages using single channel rasters.
-file_names = list.files(single_channel_dir,full.names = TRUE)
-image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
-gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
- names = image_names,
- negative_y = F,
- flip_vertical = T)
-names(gimg_list) <- image_names
-plot(gimg_list[['Vimentin']])
+file_names = list.files(single_channel_dir,full.names = TRUE)
+image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
+gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
+ names = image_names,
+ negative_y = F,
+ flip_vertical = T)
+names(gimg_list) <- image_names
+plot(gimg_list[['Vimentin']])
@@ -618,34 +618,34 @@ 11.3 Cell Segmentation file to ge
11.3.1 Using segmentation output file from DeepCell(mesmer) as an example.
We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function.
-gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
-plot(gpoly_mesmer)
+gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
+plot(gpoly_mesmer)
We can also zoom in to check how does the segmentation look.
-zoom <- c(2000,2500,2000,2500)
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+zoom <- c(2000,2500,2000,2500)
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
11.3.2 Using Giotto wrapper of Cellpose to perform segmentation
-gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
-writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
-#Create a giotto image to evaluate segmentation
-gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
+gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
+writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
+#Create a giotto image to evaluate segmentation
+gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website
-doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
- mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- channel_1 = 0,
- channel_2 = 0,
- model_name = 'cyto3',
- batch_size=12)
-cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- shift_horizontal_step = F,
- shift_vertical_step = F,
- flip_vertical = T)
-plot(gimg_for_cellpose)
-plot(cpoly, add = T, border = 'red')
+doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
+ mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ channel_1 = 0,
+ channel_2 = 0,
+ model_name = 'cyto3',
+ batch_size=12)
+cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ shift_horizontal_step = F,
+ shift_vertical_step = F,
+ flip_vertical = T)
+plot(gimg_for_cellpose)
+plot(cpoly, add = T, border = 'red')
@@ -654,133 +654,133 @@ 11.4 Create a Giotto Object using
You will need to have
- list of giotto images
- giottoPolygon created from segmentation
-Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
- images = gimg_list,
- instructions = instrs)
-Lunaphore_giotto
+Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
+ images = gimg_list,
+ instructions = instrs)
+Lunaphore_giotto
11.4.1 Overlap to matrix
calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with
-Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
- spatial_info = 'cell',
- image_names = names(gimg_list))
-
-Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
- type ='intensity',
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-showGiottoExpression(Lunaphore_giotto)
+Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
+ spatial_info = 'cell',
+ image_names = names(gimg_list))
+
+Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
+ type ='intensity',
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+showGiottoExpression(Lunaphore_giotto)
11.4.2 Manipulate Expression information
For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile.
Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object
-expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
-filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
-Lunaphore_giotto <- setExpression(Lunaphore_giotto,
- feat_type = 'protein',
- x = createExprObj(filtered_expr_mtx),
- name = 'raw')
-showGiottoExpression(Lunaphore_giotto)
+expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
+filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
+Lunaphore_giotto <- setExpression(Lunaphore_giotto,
+ feat_type = 'protein',
+ x = createExprObj(filtered_expr_mtx),
+ name = 'raw')
+showGiottoExpression(Lunaphore_giotto)
11.4.3 Rescale polygons
rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis
-Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
- poly_info = 'cell',
- name = 'smallcell',
- fx = 0.7, fy = 0.7, calculate_centroids = T)
-smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
-plot(smallpoly,add = TRUE, border = "red",ext = zoom)
+Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
+ poly_info = 'cell',
+ name = 'smallcell',
+ fx = 0.7, fy = 0.7, calculate_centroids = T)
+smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+plot(smallpoly,add = TRUE, border = "red",ext = zoom)
11.4.4 Perform clustering and differential expression
The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering
-Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
-Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
-
-Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
- scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
- set_seed = TRUE)
-screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
+Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
+Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
+
+Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
+ scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
+ set_seed = TRUE)
+screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check.
-Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
- dimensions_to_use = 1:5,
- set_seed = TRUE)
-Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
-Lunaphore_giotto = doHclust(Lunaphore_giotto,
- k = 8,
- dim_reduction_to_use = 'cells',
- spat_unit = 'cell',
- feat_type = 'protein')
-
-spatInSituPlotPoints(gobject = Lunaphore_giotto,
- spat_unit = "cell",
- polygon_feat_type = "cell",
- show_polygon = T,
- feat_type = "protein",
- feats = NULL,
- polygon_fill = 'hclust',
- polygon_fill_as_factor = TRUE,
- polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
- polygon_color = 'black',
- background_color = "white")
+Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
+ dimensions_to_use = 1:5,
+ set_seed = TRUE)
+Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
+Lunaphore_giotto = doHclust(Lunaphore_giotto,
+ k = 8,
+ dim_reduction_to_use = 'cells',
+ spat_unit = 'cell',
+ feat_type = 'protein')
+
+spatInSituPlotPoints(gobject = Lunaphore_giotto,
+ spat_unit = "cell",
+ polygon_feat_type = "cell",
+ show_polygon = T,
+ feat_type = "protein",
+ feats = NULL,
+ polygon_fill = 'hclust',
+ polygon_fill_as_factor = TRUE,
+ polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
+ polygon_color = 'black',
+ background_color = "white")
Then we can check the heatmap of protein expression and determine the first round of cluster annotation
-cluster_column = 'hclust'
-plotMetaDataHeatmap(Lunaphore_giotto,
- spat_unit = 'cell',feat_type = 'protein',
- expression_values = "raw",
- metadata_cols = c(cluster_column),
- selected_feats = names(gimg_list),
- y_text_size = 8,
- show_values = 'zscores_rescaled')
+cluster_column = 'hclust'
+plotMetaDataHeatmap(Lunaphore_giotto,
+ spat_unit = 'cell',feat_type = 'protein',
+ expression_values = "raw",
+ metadata_cols = c(cluster_column),
+ selected_feats = names(gimg_list),
+ y_text_size = 8,
+ show_values = 'zscores_rescaled')
11.4.5 Give the cluster an annotation based on expression values
-
+
11.4.6 spatial network
This is to create a cellular neighborhood based on nearest neighbor of physical distance
-Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
-
-spatPlot2D(Lunaphore_giotto,
- show_network = T,
- network_color = 'blue',
- point_size = 1.5,
- cell_color = 'hclust')
+Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
+
+spatPlot2D(Lunaphore_giotto,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1.5,
+ cell_color = 'hclust')
11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions
This is using cellProximityEnrichment() to statistically identify cell type interactions
-cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
- cluster_column = 'cell_types',
- spatial_network_name = 'Delaunay_network',
- adjust_method = 'fdr',
- number_of_simulations = 2000)
-## barplot
-cellProximityBarplot(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- min_orig_ints = 5, min_sim_ints = 5)
+cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
+ cluster_column = 'cell_types',
+ spatial_network_name = 'Delaunay_network',
+ adjust_method = 'fdr',
+ number_of_simulations = 2000)
+## barplot
+cellProximityBarplot(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ min_orig_ints = 5, min_sim_ints = 5)
-## network
-cellProximityNetwork(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- remove_self_edges = T,
- only_show_enrichment_edges = F)
+## network
+cellProximityNetwork(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ remove_self_edges = T,
+ only_show_enrichment_edges = F)
diff --git a/visium-hd.html b/visium-hd.html
index 23aa210..f24cdd4 100644
--- a/visium-hd.html
+++ b/visium-hd.html
@@ -525,7 +525,7 @@ 9.1 Objective9.2 Background
9.2.1 Visium HD Technology
-
+
Figure 9.1: Overview of Visium HD. Source: 10X Genomics
@@ -536,7 +536,7 @@
9.2.1 Visium HD Technology
9.2.2 Colorectal Cancer Sample
-
+
Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics
@@ -550,7 +550,7 @@
9.2.2 Colorectal Cancer Sample9.3 Data Ingestion
9.3.1 Visium HD output data format
-
+
Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline.
@@ -560,19 +560,19 @@
9.3.1 Visium HD output data forma
9.4 Tiling and aggregation
The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells.
-
+
9.5 Scalability and projection functions
diff --git a/visium-part-i.html b/visium-part-i.html
index 5aa1658..c4746ee 100644
--- a/visium-part-i.html
+++ b/visium-part-i.html
@@ -520,7 +520,7 @@ 7 Visium Part I
7.1 The Visium technology
Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression.
-
+
Figure 7.1: Visum workflow. Source: 10X Genomics
@@ -536,73 +536,73 @@
7.2 Introduction to the spatial d
7.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/01_session5/")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
+dir.create("data/01_session5/")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”.
-
+
7.4 Create the Giotto object
createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/01_session5/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-## Provide the path to the visium folder
-data_path <- "data/01_session5/"
-
-## Create object directly from the visium folder
-visium_brain <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/01_session5/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+## Provide the path to the visium folder
+data_path <- "data/01_session5/"
+
+## Create object directly from the visium folder
+visium_brain <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
7.5 Subset on spots that were covered by tissue
Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area.
-spatPlot2D(
- gobject = visium_brain,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE)
-
+spatPlot2D(
+ gobject = visium_brain,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE)
+
Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area.
-
+
7.6 Quality control
@@ -610,43 +610,43 @@ 7.6 Quality controlvisium_brain_statistics <- addStatistics(gobject = visium_brain,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = visium_brain_statistics,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain_statistics <- addStatistics(gobject = visium_brain,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = visium_brain_statistics,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.3: Spatial distribution of features per spot.
filterDistributions() creates a histogram to show the distribution of features per spot across the sample.
-
-
+
+
Figure 7.4: Distribution of features per spot.
When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample.
-
-
+
+
Figure 7.5: Distribution of cells with different features per spot.
filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data:
-filterCombinations(gobject = visium_brain_statistics,
- expression_thresholds = c(1, 2, 3),
- feat_det_in_min_cells = c(50, 100, 200),
- min_det_feats_per_cell = c(500, 1000, 1500))
-
+filterCombinations(gobject = visium_brain_statistics,
+ expression_thresholds = c(1, 2, 3),
+ feat_det_in_min_cells = c(50, 100, 200),
+ min_det_feats_per_cell = c(500, 1000, 1500))
+
Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations.
@@ -656,34 +656,34 @@
7.6 Quality control
7.7 Filtering
Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample.
-visium_brain <- filterGiotto(
- gobject = visium_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE
-)
-Feature type: rna
-Number of cells removed: 4 out of 2702
-Number of feats removed: 7311 out of 22125
+
+
7.8 Normalization
Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one.
-visium_brain <- normalizeGiotto(
- gobject = visium_brain,
- scalefactor = 6000,
- verbose = TRUE
-)
+visium_brain <- normalizeGiotto(
+ gobject = visium_brain,
+ scalefactor = 6000,
+ verbose = TRUE
+)
Calculate the normalized number of features per spot and save the statistics in the metadata table.
-visium_brain <- addStatistics(gobject = visium_brain)
-
-## visualize
-spatPlot2D(gobject = visium_brain,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain <- addStatistics(gobject = visium_brain)
+
+## visualize
+spatPlot2D(gobject = visium_brain,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.7: Spatial distribution of the number of features per spot.
@@ -698,11 +698,11 @@
7.9.1 Highly Variable Features:
loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_loess",
- save_plot = TRUE,
- default_save_name = "HVFplot_loess")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_loess",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_loess")
+
Figure 7.8: Covariance of HVFs using the loess method.
@@ -711,11 +711,11 @@
7.9.1 Highly Variable Features:
pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")
+
Figure 7.9: Variance of HVFs using the pearson residuals method.
@@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
Figure 17.1: Overview of Visium. Source: 10X Genomics. @@ -542,94 +542,94 @@
17.1.3 Generating a geojson file
17.2 Downloading the dataset
We first need to import a dataset that we want to perform kriging on.
-data_directory <- "data"
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz")
+data_directory <- "data"
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”.
17.3 Importing visium data
We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network.
-library(Giotto)
-
-save_directory <- 'results'
-visium_save_directory <- file.path(save_directory, 'visium_mouse_brain')
-subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/')
-
-instrs <- createGiottoInstructions(show_plot = TRUE,
- save_plot = TRUE,
- save_dir = visium_save_directory)
-
-v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs)
-
-# Subset to in tissue only
-cm = pDataDT(v_brain)
-in_tissue_barcodes = cm[in_tissue == 1]$cell_ID
-v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes)
-
-# Filter
-v_brain = filterGiotto(gobject = v_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = c('raw'))
-
-# Normalize
-v_brain = normalizeGiotto(gobject = v_brain,
- scalefactor = 6000,
- verbose = TRUE)
-
-# Add stats
-v_brain = addStatistics(gobject = v_brain)
-
-# ID HVF
-v_brain = calculateHVF(gobject = v_brain, method = "cov_loess")
-fm = fDataDT(v_brain)
-hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID
-length(hv_feats)
-
-# Dimension Reductions
-v_brain = runPCA(gobject = v_brain,
- feats_to_use = hv_feats)
-
-v_brain = runUMAP(v_brain,
- dimensions_to_use = 1:10,
- n_neighbors = 15,
- set_seed = TRUE)
-
-# NN Network
-v_brain = createNearestNetwork(gobject = v_brain,
- dimensions_to_use = 1:10,
- k = 15)
-# Leiden Cluster
-v_brain = doLeidenCluster(gobject = v_brain,
- resolution = 0.4,
- n_iterations = 1000,
- set_seed = TRUE)
-
-# Spatial Network (kNN)
-v_brain <- createSpatialNetwork(gobject = v_brain,
- method = 'kNN',
- k = 5,
- maximum_distance_knn = 400,
- name = 'spatial_network')
-
-spatPlot2D(gobject = v_brain,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 1.5,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = T,
- save_param = list(save_name = '03_ses6_1_vis_spat'))
+library(Giotto)
+
+save_directory <- 'results'
+visium_save_directory <- file.path(save_directory, 'visium_mouse_brain')
+subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/')
+
+instrs <- createGiottoInstructions(show_plot = TRUE,
+ save_plot = TRUE,
+ save_dir = visium_save_directory)
+
+v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs)
+
+# Subset to in tissue only
+cm = pDataDT(v_brain)
+in_tissue_barcodes = cm[in_tissue == 1]$cell_ID
+v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes)
+
+# Filter
+v_brain = filterGiotto(gobject = v_brain,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1000,
+ expression_values = c('raw'))
+
+# Normalize
+v_brain = normalizeGiotto(gobject = v_brain,
+ scalefactor = 6000,
+ verbose = TRUE)
+
+# Add stats
+v_brain = addStatistics(gobject = v_brain)
+
+# ID HVF
+v_brain = calculateHVF(gobject = v_brain, method = "cov_loess")
+fm = fDataDT(v_brain)
+hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID
+length(hv_feats)
+
+# Dimension Reductions
+v_brain = runPCA(gobject = v_brain,
+ feats_to_use = hv_feats)
+
+v_brain = runUMAP(v_brain,
+ dimensions_to_use = 1:10,
+ n_neighbors = 15,
+ set_seed = TRUE)
+
+# NN Network
+v_brain = createNearestNetwork(gobject = v_brain,
+ dimensions_to_use = 1:10,
+ k = 15)
+# Leiden Cluster
+v_brain = doLeidenCluster(gobject = v_brain,
+ resolution = 0.4,
+ n_iterations = 1000,
+ set_seed = TRUE)
+
+# Spatial Network (kNN)
+v_brain <- createSpatialNetwork(gobject = v_brain,
+ method = 'kNN',
+ k = 5,
+ maximum_distance_knn = 400,
+ name = 'spatial_network')
+
+spatPlot2D(gobject = v_brain,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 1.5,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = T,
+ save_param = list(save_name = '03_ses6_1_vis_spat'))
Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain.
-
+
Figure 17.2: Mouse brain spatial plot showing leiden clustering
@@ -638,17 +638,17 @@
17.3 Importing visium data
17.3.1 Identifying spatially organized features
We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time.
-# Spatially Variable Features
-ranktest = binSpect(v_brain,
- bin_method = 'rank',
- calc_hub = T,
- hub_min_int = 5,
- spatial_network_name = 'spatial_network',
- do_parallel = T,
- cores = 8) #not able to provide a seed number, so do not set one
-
-# Getting the top 15 spatially organized genes
-ext_spatial_features = ranktest[1:15,]$feats
+# Spatially Variable Features
+ranktest = binSpect(v_brain,
+ bin_method = 'rank',
+ calc_hub = T,
+ hub_min_int = 5,
+ spatial_network_name = 'spatial_network',
+ do_parallel = T,
+ cores = 8) #not able to provide a seed number, so do not set one
+
+# Getting the top 15 spatially organized genes
+ext_spatial_features = ranktest[1:15,]$feats
@@ -656,27 +656,27 @@ 17.4 Adding cell polygons to Giot
17.4.1 Read in the poly information
First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object.
-# Read in the data
-stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson")
-
-stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path,
- name = "stardist_cell",
- calc_centroids = TRUE)
-
-stardist_cell_gpoly <- flip(stardist_cell_gpoly)
+# Read in the data
+stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson")
+
+stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path,
+ name = "stardist_cell",
+ calc_centroids = TRUE)
+
+stardist_cell_gpoly <- flip(stardist_cell_gpoly)
17.4.2 Vizualizing polygons
Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain.
-
-
+
+
Figure 17.3: Mouse brain cell polygons from the visium dataset
-
+
Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped
@@ -686,8 +686,8 @@
17.4.2 Vizualizing polygons
17.4.3 Showing Giotto object prior to polygon addition
Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon.
-
-
+
+
Figure 17.5: Giotto object before adding subcellular polygons.
@@ -697,11 +697,11 @@
17.4.3 Showing Giotto object prio
17.4.4 Adding polygons to giotto object
After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject.
-v_brain <- addGiottoPolygons(v_brain,
- gpolygons = list('stardist_cell' = stardist_cell_gpoly))
-
-print(v_brain)
-
+v_brain <- addGiottoPolygons(v_brain,
+ gpolygons = list('stardist_cell' = stardist_cell_gpoly))
+
+print(v_brain)
+
Figure 17.6: Giotto object after to adding subcellular polygons.
@@ -711,9 +711,9 @@
17.4.4 Adding polygons to giotto
17.4.5 Check polygon information
We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots.
-
-
+
+
Figure 17.7: Polygon information for stardist_cell.
@@ -727,23 +727,23 @@
17.5 Performing kriging17.5.1 Interpolating features
Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be
analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required.
-future::plan(future::multisession()) # comment out for single threading
-subcell_v_brain <- interpolateFeature(v_brain,
- spat_unit = 'cell',
- feat_type = 'rna',
- ext = ext(v_brain),
- feats = ext_spatial_features,
- overwrite = T)
-
-print(subcell_v_brain)
-
+future::plan(future::multisession()) # comment out for single threading
+subcell_v_brain <- interpolateFeature(v_brain,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ ext = ext(v_brain),
+ feats = ext_spatial_features,
+ overwrite = T)
+
+print(subcell_v_brain)
+
Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right).
For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1.
-
+
Figure 17.9: Raster of gene expression interpolation for Pantr1
@@ -754,18 +754,18 @@
17.5.1 Interpolating features17.5.2 Expression overlap
The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier.
This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run.
-subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain,
- name_overlap = "rna",
- spatial_info = "stardist_cell",
- image_names = ext_spatial_features)
-
-subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain,
- poly_info = "stardist_cell",
- feat_info = "rna",
- aggr_function = "sum",
- type='intensity')
+subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain,
+ name_overlap = "rna",
+ spatial_info = "stardist_cell",
+ image_names = ext_spatial_features)
+
+subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain,
+ poly_info = "stardist_cell",
+ feat_info = "rna",
+ aggr_function = "sum",
+ type='intensity')
After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified.
-
+
Figure 17.10: Gene expression for cells based on interpolation.
@@ -776,70 +776,70 @@
17.5.2 Expression overlap
17.6 Reading in larger dataset
For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here.
-
+
17.7 Analyzing interpolated features
17.7.1 Filter and normalization
Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis.
-subcell_v_brain <- filterGiotto(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- expression_values = "raw",
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-
-subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- scalefactor = 6000,
- verbose = T)
+subcell_v_brain <- filterGiotto(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ expression_values = "raw",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+
+subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ scalefactor = 6000,
+ verbose = T)
17.7.2 Visualizing gene expression from interpolated expression
Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data.
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Sparc',
- point_size = 2,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_sparc_vis'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'stardist_cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Sparc',
- point_size = 0.6,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_sparc'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'cell',
- gradient_style = 'sequential',
- feats = 'Pantr1',
- cell_color_gradient = 'Geyser',
- point_size = 2,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_pantr1_vis'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'stardist_cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Pantr1',
- point_size = 0.6,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_pantr1'))
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Sparc',
+ point_size = 2,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_sparc_vis'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Sparc',
+ point_size = 0.6,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_sparc'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'cell',
+ gradient_style = 'sequential',
+ feats = 'Pantr1',
+ cell_color_gradient = 'Geyser',
+ point_size = 2,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_pantr1_vis'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Pantr1',
+ point_size = 0.6,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_pantr1'))
Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer.
-
+
Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom).
@@ -848,37 +848,37 @@
17.7.2 Visualizing gene expressio
17.7.3 Run PCA
-
+
17.7.4 Clustering
-# UMAP
-subcell_v_brain <- runUMAP(subcell_v_brain,
- spat_unit = "stardist_cell",
- dimensions_to_use = 1:15,
- n_neighbors = 1000,
- min_dist = 0.001,
- spread = 1)
-
-# NN Network
-subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- dimensions_to_use = 1:10,
- feats_to_use = hv_feats,
- expression_values = 'normalized',
- k = 70)
-
-subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- resolution = 0.15,
- n_iterations = 100,
- partition_type = 'RBConfigurationVertexPartition')
-
-plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus')
-
+# UMAP
+subcell_v_brain <- runUMAP(subcell_v_brain,
+ spat_unit = "stardist_cell",
+ dimensions_to_use = 1:15,
+ n_neighbors = 1000,
+ min_dist = 0.001,
+ spread = 1)
+
+# NN Network
+subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ dimensions_to_use = 1:10,
+ feats_to_use = hv_feats,
+ expression_values = 'normalized',
+ k = 70)
+
+subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ resolution = 0.15,
+ n_iterations = 100,
+ partition_type = 'RBConfigurationVertexPartition')
+
+plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus')
+
Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering.
@@ -888,27 +888,27 @@
17.7.4 Clustering
17.7.5 Visualizing clustering
Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster.
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.5,
- point_shape = 'no_border',
- background_color = 'black',
- save_plot = F,
- show_legend = TRUE)
-
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat'))
-
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.5,
+ point_shape = 'no_border',
+ background_color = 'black',
+ save_plot = F,
+ show_legend = TRUE)
+
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat'))
+
Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
@@ -918,37 +918,37 @@
17.7.5 Visualizing clustering
17.7.6 Cropping objects
We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below.
-subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
- spat_unit = ":all:",
- x_min = 4000,
- x_max = 7000,
- y_min = -6500,
- y_max = -3500,
- z_max = NULL,
- z_min = NULL)
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 2,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_vis_spat_crop'))
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat_crop'))
-
+subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
+ spat_unit = ":all:",
+ x_min = 4000,
+ x_max = 7000,
+ y_min = -6500,
+ y_max = -3500,
+ z_max = NULL,
+ z_min = NULL)
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 2,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_vis_spat_crop'))
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat_crop'))
+
Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
diff --git a/interoperability-with-isolated-tools.html b/interoperability-with-isolated-tools.html
index ceb2d0a..8cca7e6 100644
--- a/interoperability-with-isolated-tools.html
+++ b/interoperability-with-isolated-tools.html
@@ -526,1521 +526,1521 @@
16.1.1 Dataset downloaddata_path <- "data"
-dir.create(data_path)
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
- destfile = file.path(data_path,"counts.h5ad"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
- destfile = file.path(data_path,"cell_labels.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
+data_path <- "data"
+dir.create(data_path)
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
+ destfile = file.path(data_path,"counts.h5ad"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
+ destfile = file.path(data_path,"cell_labels.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
16.1.2 Create the Giotto object
-library(Giotto)
-library(reticulate)
-
-## Set instructions
-results_folder <- "results"
-dir.create(results_folder)
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-
-## load data
-
-### meta data
-meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-colnames(meta_df)[[1]] <- "cell_ID"
-slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
-slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
-
-### counts matrix
-scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
-ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
-reticulate::source_python(ad2g_path)
-adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
-X <- extract_expression(adata)
-cID <- extract_cell_IDs(adata)
-fID <- extract_feat_IDs(adata)
-rownames(X) <- fID
-colnames(X) <- cID
-slice1_filter <- cID %in% slice1_meta_df$cell_ID
-slice2_filter <- cID %in% slice2_meta_df$cell_ID
-slice1_exp <- X[,slice1_filter]
-slice2_exp <- X[,slice2_filter]
-
-### cell segmentation to cell location
-segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_df <- rbind(segments_1_df, segments_2_df)
-loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
-loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
-loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
-centr.x <- apply(loc.use[,loc.x],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-centr.y <- apply(loc.use[,loc.y],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
-slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
-slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
-
-## giotto obj
-giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
- spatial_locs = slice1_spatial_locs_df,
- cell_metadata = slice1_meta_df)
-giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
- spatial_locs = slice2_spatial_locs_df,
- cell_metadata = slice2_meta_df)
-giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
- gobject_names = c('mouse2_slice229',
- 'mouse2_slice300'), # name for each samples
- join_method = 'z_stack')
-# We skip the processing process here to save time and use the given cell type
-# annotation directly
-
-ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
- cell_type = "subclass",
- output_path = data_path,
- spat_unit = "cell",
- feat_type = "rna",
- verbose = TRUE)
-
-# Cell_ID Sample x y Cell_Type
-# <char> <char> <dbl> <dbl> <char>
-# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
-# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
-# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
-# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
-# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
-# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
+library(Giotto)
+library(reticulate)
+
+## Set instructions
+results_folder <- "results"
+dir.create(results_folder)
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+
+## load data
+
+### meta data
+meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+colnames(meta_df)[[1]] <- "cell_ID"
+slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
+slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
+
+### counts matrix
+scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
+ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
+reticulate::source_python(ad2g_path)
+adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
+X <- extract_expression(adata)
+cID <- extract_cell_IDs(adata)
+fID <- extract_feat_IDs(adata)
+rownames(X) <- fID
+colnames(X) <- cID
+slice1_filter <- cID %in% slice1_meta_df$cell_ID
+slice2_filter <- cID %in% slice2_meta_df$cell_ID
+slice1_exp <- X[,slice1_filter]
+slice2_exp <- X[,slice2_filter]
+
+### cell segmentation to cell location
+segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_df <- rbind(segments_1_df, segments_2_df)
+loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
+loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
+loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
+centr.x <- apply(loc.use[,loc.x],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+centr.y <- apply(loc.use[,loc.y],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
+slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
+slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
+
+## giotto obj
+giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
+ spatial_locs = slice1_spatial_locs_df,
+ cell_metadata = slice1_meta_df)
+giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
+ spatial_locs = slice2_spatial_locs_df,
+ cell_metadata = slice2_meta_df)
+giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
+ gobject_names = c('mouse2_slice229',
+ 'mouse2_slice300'), # name for each samples
+ join_method = 'z_stack')
+# We skip the processing process here to save time and use the given cell type
+# annotation directly
+
+ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
+ cell_type = "subclass",
+ output_path = data_path,
+ spat_unit = "cell",
+ feat_type = "rna",
+ verbose = TRUE)
+
+# Cell_ID Sample x y Cell_Type
+# <char> <char> <dbl> <dbl> <char>
+# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
+# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
+# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
+# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
+# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
+# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
Spatial cell type distributions
-spatPlot2D(giotto_obj,
- group_by = "slice_id",
- cell_color = "subclass",
- point_size = 1,
- point_border_stroke = NA,
- legend_text = 6)
+spatPlot2D(giotto_obj,
+ group_by = "slice_id",
+ cell_color = "subclass",
+ point_size = 1,
+ point_border_stroke = NA,
+ legend_text = 6)
16.1.2.1 Installation ONTraC
You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node.
It will run for less than 10 minutes on this example dataset.
For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time.
-source ~/.bash_profile
-conda create -y -n ONTraC python=3.11
-conda activate ONTraC
-pip install git+https://github.com/gyuanlab/ONTraC.git@V2
-# Retrieving notices: ...working... done
-# Channels:
-# - conda-forge
-# - bioconda
-# - r
-# - pytorch
-# - defaults
-# Platform: osx-arm64
-# Collecting package metadata (repodata.json): ...working... done
-# Solving environment: ...working... done
-#
-# ## Package Plan ##
-#
-# environment location: /Users/wwang/miniconda3/envs/ONTraC
-#
-# added / updated specs:
-# - python=3.11
-#
-#
-# The following packages will be downloaded:
-#
-# package | build
-# ---------------------------|-----------------
-# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
-# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
-# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
-# ------------------------------------------------------------
-# Total: 4.3 MB
-#
-# The following NEW packages will be INSTALLED:
-#
-# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
-# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
-# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
-# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
-# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
-# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
-# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
-# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
-# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
-# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
-# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
-# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
-# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
-# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
-# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
-# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
-#
-#
-#
-# Downloading and Extracting Packages: ...working... done
-# Preparing transaction: ...working... done
-# Verifying transaction: ...working... done
-# Executing transaction: ...working... done
-# #
-# # To activate this environment, use
-# #
-# # $ conda activate ONTraC
-# #
-# # To deactivate an active environment, use
-# #
-# # $ conda deactivate
-#
-# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
-# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
-# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
-# Running command git checkout -b V2 --track origin/V2
-# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
-# Installing build dependencies: started
-# Switched to a new branch 'V2'
-# branch 'V2' set up to track 'origin/V2'.
-# Installing build dependencies: finished with status 'done'
-# Getting requirements to build wheel: started
-# Getting requirements to build wheel: finished with status 'done'
-# Preparing metadata (pyproject.toml): started
-# Preparing metadata (pyproject.toml): finished with status 'done'
-# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
-# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
-# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
-# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
-# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
-# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
-# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
-# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
-# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
-# Collecting session-info (from ONTraC==2.0a9.dev7)
-# Using cached session_info-1.0.0-py3-none-any.whl
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
-# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
-# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
-# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
-# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
-# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
-# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
-# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
-# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
-# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
-# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
-# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
-# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
-# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
-# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
-# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
-# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
-# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
-# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
-# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
-# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
-# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
-# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
-# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
-# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
-# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
-# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
-# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
-# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
-# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
-# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
-# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
-# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
-# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
-# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
-# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
-# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
-# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
-# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
-# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
-# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
-# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
-# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
-# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
-# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
-# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
-# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
-# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
-# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
-# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
-# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
-# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
-# Using cached idna-3.7-py3-none-any.whl (66 kB)
-# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
-# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
-# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
-# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
-# Building wheels for collected packages: ONTraC
-# Building wheel for ONTraC (pyproject.toml): started
-# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
-# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
-# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
-# Successfully built ONTraC
-# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
-# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
+source ~/.bash_profile
+conda create -y -n ONTraC python=3.11
+conda activate ONTraC
+pip install git+https://github.com/gyuanlab/ONTraC.git@V2
+# Retrieving notices: ...working... done
+# Channels:
+# - conda-forge
+# - bioconda
+# - r
+# - pytorch
+# - defaults
+# Platform: osx-arm64
+# Collecting package metadata (repodata.json): ...working... done
+# Solving environment: ...working... done
+#
+# ## Package Plan ##
+#
+# environment location: /Users/wwang/miniconda3/envs/ONTraC
+#
+# added / updated specs:
+# - python=3.11
+#
+#
+# The following packages will be downloaded:
+#
+# package | build
+# ---------------------------|-----------------
+# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
+# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
+# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
+# ------------------------------------------------------------
+# Total: 4.3 MB
+#
+# The following NEW packages will be INSTALLED:
+#
+# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
+# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
+# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
+# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
+# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
+# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
+# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
+# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
+# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
+# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
+# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
+# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
+# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
+# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
+# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
+# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
+#
+#
+#
+# Downloading and Extracting Packages: ...working... done
+# Preparing transaction: ...working... done
+# Verifying transaction: ...working... done
+# Executing transaction: ...working... done
+# #
+# # To activate this environment, use
+# #
+# # $ conda activate ONTraC
+# #
+# # To deactivate an active environment, use
+# #
+# # $ conda deactivate
+#
+# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
+# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
+# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
+# Running command git checkout -b V2 --track origin/V2
+# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
+# Installing build dependencies: started
+# Switched to a new branch 'V2'
+# branch 'V2' set up to track 'origin/V2'.
+# Installing build dependencies: finished with status 'done'
+# Getting requirements to build wheel: started
+# Getting requirements to build wheel: finished with status 'done'
+# Preparing metadata (pyproject.toml): started
+# Preparing metadata (pyproject.toml): finished with status 'done'
+# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
+# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
+# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
+# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
+# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
+# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
+# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
+# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
+# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
+# Collecting session-info (from ONTraC==2.0a9.dev7)
+# Using cached session_info-1.0.0-py3-none-any.whl
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
+# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
+# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
+# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
+# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
+# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
+# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
+# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
+# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
+# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
+# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
+# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
+# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
+# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
+# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
+# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
+# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
+# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
+# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
+# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
+# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
+# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
+# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
+# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
+# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
+# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
+# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
+# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
+# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
+# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
+# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
+# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
+# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
+# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
+# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
+# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
+# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
+# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
+# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
+# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
+# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
+# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
+# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
+# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
+# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
+# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
+# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
+# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
+# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
+# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
+# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
+# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
+# Using cached idna-3.7-py3-none-any.whl (66 kB)
+# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
+# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
+# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
+# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
+# Building wheels for collected packages: ONTraC
+# Building wheel for ONTraC (pyproject.toml): started
+# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
+# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
+# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
+# Successfully built ONTraC
+# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
+# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
16.1.2.2 Running ONTraC
-
-source ~/.bash_profile
-conda activate ONTraC_v2_py311
-ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
-# ##################################################################################
-#
-# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
-# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
-# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
-# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
-# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
-#
-# version: 2.0a11
-#
-# ##################################################################################
-# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
-# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
-# 11:33:23 --- INFO: -------- I/O options -------
-# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
-# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
-# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
-# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
-# 11:33:23 --- INFO: -------- preprocessing options -------
-# 11:33:23 --- INFO: resolution: 10.0
-# 11:33:23 --- INFO: -------- niche net constr options -------
-# 11:33:23 --- INFO: n_cpu: 4
-# 11:33:23 --- INFO: n_neighbors: 50
-# 11:33:23 --- INFO: n_local: 20
-# 11:33:23 --- INFO: embedding_adjust: False
-# 11:33:23 --- INFO: sigma: 1
-# 11:33:23 --- INFO: -------- train options -------
-# 11:33:23 --- INFO: device: cpu
-# 11:33:23 --- INFO: epochs: 1000
-# 11:33:23 --- INFO: batch_size: 0
-# 11:33:23 --- INFO: patience: 100
-# 11:33:23 --- INFO: min_delta: 0.001
-# 11:33:23 --- INFO: min_epochs: 50
-# 11:33:23 --- INFO: seed: 42
-# 11:33:23 --- INFO: lr: 0.03
-# 11:33:23 --- INFO: hidden_feats: 4
-# 11:33:23 --- INFO: k: 6
-# 11:33:23 --- INFO: modularity_loss_weight: 1.0
-# 11:33:23 --- INFO: purity_loss_weight: 100.0
-# 11:33:23 --- INFO: regularization_loss_weight: 0.1
-# 11:33:23 --- INFO: beta: 0.3
-# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
-# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
-# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
-# 11:33:23 --- INFO: ------------- Niche network construct ---------------
-# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
-# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
-# 11:33:28 --- INFO: Generating samples.yaml file.
-# 11:33:28 --- INFO: ------------ Niche network construct end ------------
-# 11:33:28 --- INFO: ------------------------ GNN ------------------------
-# 11:33:28 --- INFO: Loading dataset.
-# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
-# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
-# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
-# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
-# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
-# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
-# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
-# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
-# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
-# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
-# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
-# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
-# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
-# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
-# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
-# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
-# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
-# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
-# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
-# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
-# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
-# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
-# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
-# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
-# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
-# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
-# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
-# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
-# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
-# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
-# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
-# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
-# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
-# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
-# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
-# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
-# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
-# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
-# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
-# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
-# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
-# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
-# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
-# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
-# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
-# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
-# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
-# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
-# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
-# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
-# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
-# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
-# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
-# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
-# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
-# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
-# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
-# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
-# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
-# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
-# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
-# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
-# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
-# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
-# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
-# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
-# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
-# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
-# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
-# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
-# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
-# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
-# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
-# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
-# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
-# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
-# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
-# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
-# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
-# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
-# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
-# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
-# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
-# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
-# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
-# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
-# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
-# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
-# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
-# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
-# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
-# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
-# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
-# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
-# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
-# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
-# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
-# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
-# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
-# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
-# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
-# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
-# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
-# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
-# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
-# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
-# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
-# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
-# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
-# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
-# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
-# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
-# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
-# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
-# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
-# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
-# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
-# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
-# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
-# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
-# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
-# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
-# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
-# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
-# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
-# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
-# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
-# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
-# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
-# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
-# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
-# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
-# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
-# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
-# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
-# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
-# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
-# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
-# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
-# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
-# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
-# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
-# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
-# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
-# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
-# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
-# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
-# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
-# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
-# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
-# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
-# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
-# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
-# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
-# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
-# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
-# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
-# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
-# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
-# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
-# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
-# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
-# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
-# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
-# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
-# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
-# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
-# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
-# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
-# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
-# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
-# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
-# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
-# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
-# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
-# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
-# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
-# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
-# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
-# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
-# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
-# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
-# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
-# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
-# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
-# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
-# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
-# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
-# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
-# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
-# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
-# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
-# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
-# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
-# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
-# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
-# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
-# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
-# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
-# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
-# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
-# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
-# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
-# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
-# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
-# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
-# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
-# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
-# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
-# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
-# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
-# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
-# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
-# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
-# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
-# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
-# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
-# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
-# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
-# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
-# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
-# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
-# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
-# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
-# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
-# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
-# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
-# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
-# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
-# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
-# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
-# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
-# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
-# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
-# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
-# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
-# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
-# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
-# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
-# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
-# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
-# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
-# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
-# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
-# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
-# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
-# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
-# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
-# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
-# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
-# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
-# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
-# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
-# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
-# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
-# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
-# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
-# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
-# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
-# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
-# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
-# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
-# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
-# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
-# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
-# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
-# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
-# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
-# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
-# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
-# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
-# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
-# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
-# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
-# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
-# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
-# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
-# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
-# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
-# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
-# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
-# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
-# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
-# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
-# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
-# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
-# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
-# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
-# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
-# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
-# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
-# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
-# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
-# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
-# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
-# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
-# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
-# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
-# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
-# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
-# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
-# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
-# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
-# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
-# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
-# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
-# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
-# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
-# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
-# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
-# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
-# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
-# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
-# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
-# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
-# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
-# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
-# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
-# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
-# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
-# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
-# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
-# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
-# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
-# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
-# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
-# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
-# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
-# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
-# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
-# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
-# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
-# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
-# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
-# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
-# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
-# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
-# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
-# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
-# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
-# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
-# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
-# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
-# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
-# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
-# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
-# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
-# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
-# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
-# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
-# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
-# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
-# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
-# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
-# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
-# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
-# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
-# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
-# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
-# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
-# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
-# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
-# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
-# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
-# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
-# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
-# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
-# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
-# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
-# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
-# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
-# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
-# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
-# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
-# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
-# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
-# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
-# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
-# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
-# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
-# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
-# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
-# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
-# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
-# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
-# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
-# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
-# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
-# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
-# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
-# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
-# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
-# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
-# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
-# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
-# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
-# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
-# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
-# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
-# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
-# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
-# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
-# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
-# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
-# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
-# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
-# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
-# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
-# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
-# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
-# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
-# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
-# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
-# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
-# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
-# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
-# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
-# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
-# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
-# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
-# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
-# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
-# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
-# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
-# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
-# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
-# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
-# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
-# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
-# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
-# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
-# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
-# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
-# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
-# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
-# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
-# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
-# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
-# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
-# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
-# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
-# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
-# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
-# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
-# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
-# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
-# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
-# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
-# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
-# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
-# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
-# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
-# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
-# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
-# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
-# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
-# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
-# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
-# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
-# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
-# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
-# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
-# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
-# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
-# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
-# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
-# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
-# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
-# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
-# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
-# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
-# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
-# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
-# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
-# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
-# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
-# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
-# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
-# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
-# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
-# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
-# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
-# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
-# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
-# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
-# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
-# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
-# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
-# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
-# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
-# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
-# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
-# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
-# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
-# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
-# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
-# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
-# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
-# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
-# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
-# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
-# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
-# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
-# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
-# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
-# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
-# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
-# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
-# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
-# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
-# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
-# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
-# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
-# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
-# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
-# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
-# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
-# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
-# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
-# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
-# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
-# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
-# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
-# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
-# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
-# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
-# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
-# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
-# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
-# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
-# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
-# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
-# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
-# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
-# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
-# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
-# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
-# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
-# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
-# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
-# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
-# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
-# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
-# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
-# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
-# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
-# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
-# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
-# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
-# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
-# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
-# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
-# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
-# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
-# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
-# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
-# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
-# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
-# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
-# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
-# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
-# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
-# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
-# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
-# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
-# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
-# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
-# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
-# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
-# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
-# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
-# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
-# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
-# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
-# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
-# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
-# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
-# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
-# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
-# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
-# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
-# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
-# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
-# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
-# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
-# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
-# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
-# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
-# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
-# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
-# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
-# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
-# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
-# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
-# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
-# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
-# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
-# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
-# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
-# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
-# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
-# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
-# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
-# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
-# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
-# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
-# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
-# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
-# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
-# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
-# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
-# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
-# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
-# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
-# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
-# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
-# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
-# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
-# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
-# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
-# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
-# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
-# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
-# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
-# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
-# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
-# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
-# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
-# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
-# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
-# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
-# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
-# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
-# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
-# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
-# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
-# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
-# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
-# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
-# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
-# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
-# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
-# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
-# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
-# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
-# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
-# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
-# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
-# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
-# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
-# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
-# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
-# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
-# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
-# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
-# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
-# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
-# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
-# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
-# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
-# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
-# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
-# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
-# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
-# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
-# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
-# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
-# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
-# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
-# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
-# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
-# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
-# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
-# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
-# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
-# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
-# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
-# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
-# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
-# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
-# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
-# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
-# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
-# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
-# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
-# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
-# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
-# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
-# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
-# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
-# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
-# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
-# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
-# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
-# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
-# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
-# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
-# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
-# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
-# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
-# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
-# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
-# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
-# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
-# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
-# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
-# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
-# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
-# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
-# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
-# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
-# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
-# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
-# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
-# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
-# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
-# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
-# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
-# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
-# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
-# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
-# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
-# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
-# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
-# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
-# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
-# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
-# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
-# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
-# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
-# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
-# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
-# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
-# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
-# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
-# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
-# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
-# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
-# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
-# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
-# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
-# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
-# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
-# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
-# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
-# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
-# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
-# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
-# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
-# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
-# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
-# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
-# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
-# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
-# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
-# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
-# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
-# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
-# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
-# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
-# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
-# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
-# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
-# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
-# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
-# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
-# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
-# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
-# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
-# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
-# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
-# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
-# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
-# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
-# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
-# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
-# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
-# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
-# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
-# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
-# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
-# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
-# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
-# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
-# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
-# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
-# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
-# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
-# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
-# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
-# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
-# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
-# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
-# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
-# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
-# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
-# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
-# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
-# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
-# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
-# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
-# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
-# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
-# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
-# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
-# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
-# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
-# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
-# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
-# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
-# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
-# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
-# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
-# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
-# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
-# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
-# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
-# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
-# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
-# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
-# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
-# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
-# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
-# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
-# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
-# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
-# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
-# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
-# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
-# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
-# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
-# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
-# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
-# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
-# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
-# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
-# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
-# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
-# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
-# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
-# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
-# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
-# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
-# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
-# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
-# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
-# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
-# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
-# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
-# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
-# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
-# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
-# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
-# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
-# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
-# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
-# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
-# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
-# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
-# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
-# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
-# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
-# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
-# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
-# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
-# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
-# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
-# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
-# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
-# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
-# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
-# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
-# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
-# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
-# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
-# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
-# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
-# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
-# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
-# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
-# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
-# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
-# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
-# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
-# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
-# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
-# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
-# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
-# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
-# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
-# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
-# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
-# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
-# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
-# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
-# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
-# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
-# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
-# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
-# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
-# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
-# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
-# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
-# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
-# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
-# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
-# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
-# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
-# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
-# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
-# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
-# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
-# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
-# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
-# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
-# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
-# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
-# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
-# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
-# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
-# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
-# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
-# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
-# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
-# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
-# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
-# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
-# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
-# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
-# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
-# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
-# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
-# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
-# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
-# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
-# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
-# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
-# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
-# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
-# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
-# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
-# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
-# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
-# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
-# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
-# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
-# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
-# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
-# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
-# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
-# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
-# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
-# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
-# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
-# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
-# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
-# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
-# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
-# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
-# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
-# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
-# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
-# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
-# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
-# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
-# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
-# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
-# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
-# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
-# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
-# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
-# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
-# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
-# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
-# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
-# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
-# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
-# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
-# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
-# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
-# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
-# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
-# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
-# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
-# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
-# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
-# 11:38:37 --- INFO: Training process end.
-# 11:38:37 --- INFO: Evaluating process start.
-# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
-# 11:38:37 --- INFO: Evaluating process end.
-# 11:38:37 --- INFO: Predicting process start.
-# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
-# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
-# 11:38:38 --- INFO: Predicting process end.
-# 11:38:38 --- INFO: --------------------- GNN end ----------------------
-# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
-# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
-# 11:38:38 --- INFO: Loading dataset.
-# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:38:39 --- INFO: Calculating NTScore for each niche.
-# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
-# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
-# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
-# 11:38:39 --- INFO: Output NTScore tables.
-# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
+
+source ~/.bash_profile
+conda activate ONTraC_v2_py311
+ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
+# ##################################################################################
+#
+# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
+# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
+# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
+# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
+# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
+#
+# version: 2.0a11
+#
+# ##################################################################################
+# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
+# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
+# 11:33:23 --- INFO: -------- I/O options -------
+# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
+# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
+# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
+# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
+# 11:33:23 --- INFO: -------- preprocessing options -------
+# 11:33:23 --- INFO: resolution: 10.0
+# 11:33:23 --- INFO: -------- niche net constr options -------
+# 11:33:23 --- INFO: n_cpu: 4
+# 11:33:23 --- INFO: n_neighbors: 50
+# 11:33:23 --- INFO: n_local: 20
+# 11:33:23 --- INFO: embedding_adjust: False
+# 11:33:23 --- INFO: sigma: 1
+# 11:33:23 --- INFO: -------- train options -------
+# 11:33:23 --- INFO: device: cpu
+# 11:33:23 --- INFO: epochs: 1000
+# 11:33:23 --- INFO: batch_size: 0
+# 11:33:23 --- INFO: patience: 100
+# 11:33:23 --- INFO: min_delta: 0.001
+# 11:33:23 --- INFO: min_epochs: 50
+# 11:33:23 --- INFO: seed: 42
+# 11:33:23 --- INFO: lr: 0.03
+# 11:33:23 --- INFO: hidden_feats: 4
+# 11:33:23 --- INFO: k: 6
+# 11:33:23 --- INFO: modularity_loss_weight: 1.0
+# 11:33:23 --- INFO: purity_loss_weight: 100.0
+# 11:33:23 --- INFO: regularization_loss_weight: 0.1
+# 11:33:23 --- INFO: beta: 0.3
+# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
+# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
+# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
+# 11:33:23 --- INFO: ------------- Niche network construct ---------------
+# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
+# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
+# 11:33:28 --- INFO: Generating samples.yaml file.
+# 11:33:28 --- INFO: ------------ Niche network construct end ------------
+# 11:33:28 --- INFO: ------------------------ GNN ------------------------
+# 11:33:28 --- INFO: Loading dataset.
+# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
+# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
+# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
+# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
+# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
+# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
+# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
+# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
+# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
+# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
+# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
+# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
+# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
+# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
+# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
+# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
+# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
+# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
+# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
+# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
+# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
+# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
+# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
+# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
+# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
+# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
+# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
+# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
+# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
+# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
+# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
+# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
+# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
+# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
+# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
+# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
+# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
+# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
+# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
+# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
+# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
+# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
+# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
+# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
+# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
+# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
+# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
+# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
+# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
+# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
+# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
+# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
+# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
+# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
+# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
+# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
+# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
+# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
+# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
+# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
+# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
+# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
+# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
+# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
+# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
+# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
+# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
+# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
+# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
+# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
+# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
+# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
+# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
+# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
+# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
+# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
+# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
+# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
+# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
+# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
+# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
+# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
+# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
+# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
+# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
+# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
+# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
+# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
+# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
+# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
+# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
+# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
+# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
+# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
+# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
+# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
+# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
+# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
+# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
+# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
+# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
+# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
+# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
+# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
+# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
+# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
+# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
+# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
+# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
+# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
+# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
+# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
+# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
+# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
+# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
+# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
+# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
+# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
+# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
+# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
+# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
+# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
+# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
+# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
+# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
+# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
+# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
+# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
+# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
+# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
+# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
+# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
+# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
+# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
+# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
+# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
+# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
+# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
+# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
+# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
+# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
+# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
+# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
+# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
+# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
+# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
+# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
+# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
+# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
+# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
+# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
+# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
+# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
+# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
+# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
+# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
+# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
+# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
+# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
+# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
+# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
+# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
+# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
+# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
+# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
+# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
+# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
+# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
+# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
+# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
+# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
+# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
+# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
+# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
+# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
+# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
+# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
+# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
+# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
+# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
+# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
+# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
+# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
+# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
+# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
+# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
+# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
+# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
+# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
+# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
+# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
+# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
+# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
+# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
+# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
+# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
+# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
+# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
+# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
+# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
+# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
+# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
+# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
+# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
+# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
+# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
+# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
+# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
+# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
+# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
+# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
+# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
+# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
+# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
+# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
+# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
+# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
+# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
+# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
+# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
+# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
+# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
+# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
+# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
+# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
+# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
+# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
+# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
+# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
+# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
+# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
+# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
+# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
+# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
+# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
+# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
+# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
+# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
+# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
+# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
+# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
+# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
+# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
+# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
+# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
+# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
+# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
+# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
+# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
+# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
+# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
+# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
+# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
+# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
+# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
+# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
+# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
+# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
+# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
+# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
+# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
+# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
+# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
+# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
+# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
+# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
+# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
+# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
+# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
+# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
+# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
+# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
+# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
+# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
+# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
+# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
+# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
+# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
+# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
+# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
+# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
+# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
+# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
+# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
+# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
+# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
+# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
+# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
+# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
+# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
+# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
+# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
+# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
+# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
+# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
+# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
+# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
+# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
+# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
+# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
+# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
+# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
+# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
+# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
+# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
+# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
+# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
+# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
+# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
+# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
+# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
+# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
+# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
+# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
+# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
+# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
+# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
+# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
+# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
+# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
+# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
+# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
+# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
+# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
+# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
+# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
+# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
+# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
+# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
+# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
+# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
+# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
+# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
+# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
+# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
+# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
+# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
+# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
+# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
+# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
+# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
+# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
+# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
+# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
+# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
+# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
+# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
+# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
+# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
+# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
+# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
+# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
+# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
+# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
+# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
+# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
+# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
+# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
+# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
+# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
+# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
+# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
+# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
+# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
+# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
+# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
+# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
+# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
+# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
+# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
+# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
+# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
+# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
+# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
+# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
+# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
+# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
+# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
+# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
+# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
+# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
+# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
+# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
+# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
+# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
+# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
+# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
+# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
+# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
+# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
+# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
+# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
+# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
+# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
+# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
+# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
+# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
+# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
+# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
+# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
+# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
+# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
+# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
+# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
+# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
+# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
+# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
+# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
+# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
+# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
+# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
+# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
+# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
+# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
+# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
+# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
+# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
+# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
+# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
+# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
+# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
+# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
+# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
+# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
+# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
+# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
+# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
+# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
+# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
+# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
+# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
+# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
+# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
+# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
+# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
+# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
+# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
+# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
+# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
+# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
+# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
+# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
+# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
+# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
+# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
+# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
+# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
+# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
+# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
+# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
+# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
+# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
+# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
+# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
+# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
+# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
+# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
+# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
+# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
+# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
+# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
+# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
+# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
+# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
+# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
+# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
+# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
+# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
+# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
+# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
+# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
+# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
+# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
+# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
+# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
+# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
+# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
+# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
+# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
+# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
+# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
+# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
+# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
+# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
+# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
+# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
+# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
+# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
+# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
+# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
+# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
+# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
+# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
+# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
+# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
+# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
+# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
+# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
+# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
+# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
+# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
+# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
+# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
+# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
+# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
+# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
+# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
+# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
+# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
+# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
+# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
+# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
+# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
+# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
+# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
+# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
+# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
+# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
+# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
+# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
+# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
+# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
+# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
+# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
+# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
+# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
+# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
+# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
+# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
+# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
+# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
+# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
+# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
+# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
+# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
+# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
+# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
+# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
+# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
+# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
+# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
+# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
+# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
+# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
+# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
+# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
+# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
+# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
+# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
+# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
+# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
+# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
+# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
+# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
+# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
+# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
+# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
+# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
+# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
+# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
+# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
+# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
+# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
+# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
+# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
+# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
+# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
+# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
+# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
+# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
+# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
+# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
+# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
+# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
+# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
+# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
+# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
+# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
+# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
+# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
+# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
+# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
+# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
+# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
+# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
+# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
+# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
+# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
+# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
+# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
+# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
+# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
+# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
+# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
+# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
+# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
+# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
+# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
+# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
+# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
+# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
+# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
+# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
+# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
+# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
+# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
+# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
+# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
+# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
+# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
+# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
+# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
+# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
+# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
+# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
+# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
+# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
+# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
+# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
+# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
+# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
+# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
+# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
+# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
+# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
+# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
+# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
+# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
+# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
+# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
+# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
+# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
+# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
+# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
+# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
+# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
+# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
+# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
+# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
+# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
+# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
+# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
+# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
+# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
+# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
+# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
+# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
+# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
+# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
+# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
+# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
+# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
+# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
+# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
+# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
+# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
+# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
+# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
+# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
+# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
+# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
+# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
+# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
+# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
+# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
+# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
+# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
+# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
+# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
+# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
+# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
+# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
+# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
+# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
+# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
+# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
+# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
+# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
+# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
+# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
+# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
+# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
+# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
+# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
+# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
+# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
+# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
+# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
+# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
+# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
+# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
+# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
+# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
+# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
+# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
+# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
+# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
+# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
+# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
+# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
+# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
+# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
+# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
+# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
+# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
+# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
+# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
+# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
+# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
+# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
+# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
+# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
+# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
+# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
+# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
+# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
+# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
+# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
+# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
+# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
+# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
+# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
+# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
+# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
+# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
+# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
+# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
+# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
+# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
+# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
+# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
+# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
+# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
+# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
+# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
+# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
+# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
+# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
+# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
+# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
+# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
+# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
+# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
+# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
+# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
+# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
+# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
+# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
+# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
+# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
+# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
+# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
+# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
+# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
+# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
+# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
+# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
+# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
+# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
+# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
+# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
+# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
+# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
+# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
+# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
+# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
+# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
+# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
+# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
+# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
+# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
+# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
+# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
+# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
+# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
+# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
+# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
+# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
+# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
+# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
+# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
+# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
+# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
+# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
+# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
+# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
+# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
+# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
+# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
+# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
+# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
+# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
+# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
+# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
+# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
+# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
+# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
+# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
+# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
+# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
+# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
+# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
+# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
+# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
+# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
+# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
+# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
+# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
+# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
+# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
+# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
+# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
+# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
+# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
+# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
+# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
+# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
+# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
+# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
+# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
+# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
+# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
+# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
+# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
+# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
+# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
+# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
+# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
+# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
+# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
+# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
+# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
+# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
+# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
+# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
+# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
+# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
+# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
+# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
+# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
+# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
+# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
+# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
+# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
+# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
+# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
+# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
+# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
+# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
+# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
+# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
+# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
+# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
+# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
+# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
+# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
+# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
+# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
+# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
+# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
+# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
+# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
+# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
+# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
+# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
+# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
+# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
+# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
+# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
+# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
+# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
+# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
+# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
+# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
+# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
+# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
+# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
+# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
+# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
+# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
+# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
+# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
+# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
+# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
+# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
+# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
+# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
+# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
+# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
+# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
+# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
+# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
+# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
+# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
+# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
+# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
+# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
+# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
+# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
+# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
+# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
+# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
+# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
+# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
+# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
+# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
+# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
+# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
+# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
+# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
+# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
+# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
+# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
+# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
+# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
+# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
+# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
+# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
+# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
+# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
+# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
+# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
+# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
+# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
+# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
+# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
+# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
+# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
+# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
+# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
+# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
+# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
+# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
+# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
+# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
+# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
+# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
+# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
+# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
+# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
+# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
+# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
+# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
+# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
+# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
+# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
+# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
+# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
+# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
+# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
+# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
+# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
+# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
+# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
+# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
+# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
+# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
+# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
+# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
+# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
+# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
+# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
+# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
+# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
+# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
+# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
+# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
+# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
+# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
+# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
+# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
+# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
+# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
+# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
+# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
+# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
+# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
+# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
+# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
+# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
+# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
+# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
+# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
+# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
+# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
+# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
+# 11:38:37 --- INFO: Training process end.
+# 11:38:37 --- INFO: Evaluating process start.
+# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
+# 11:38:37 --- INFO: Evaluating process end.
+# 11:38:37 --- INFO: Predicting process start.
+# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
+# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
+# 11:38:38 --- INFO: Predicting process end.
+# 11:38:38 --- INFO: --------------------- GNN end ----------------------
+# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
+# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
+# 11:38:38 --- INFO: Loading dataset.
+# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:38:39 --- INFO: Calculating NTScore for each niche.
+# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
+# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
+# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
+# 11:38:39 --- INFO: Output NTScore tables.
+# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
16.1.3 visualization
16.1.3.1 load ONTraC results
-
+
The NTScore and binarized niche cluster info were stored in cell metadata
-
-# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
-# <char> <char> <char> <char> <char> <char> <char> <int> <num>
-# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
-# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
-# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
-# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
-# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
-# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
+
+# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
+# <char> <char> <char> <char> <char> <char> <char> <int> <num>
+# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
+# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
+# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
+# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
+# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
+# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
The probability matrix of each cell assigned to each niche cluster and
connectivity between niche cluster were stored here.
-
-
+
+
16.1.3.2 niche cluster prob
-spatFeatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'niche cluster',
- expression_values = "prob",
- group_by = 'list_ID',
- feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
- point_border_col = 'gray'
- )
+spatFeatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'niche cluster',
+ expression_values = "prob",
+ group_by = 'list_ID',
+ feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
+ point_border_col = 'gray'
+ )
16.1.3.3 binarized niche cluster
-spatPlot2D(giotto_obj,
- spat_unit = "cell",
- group_by = 'list_ID',
- cell_color = "NicheCluster",
- color_as_factor = T,
- point_size = 1,
- point_border_stroke = NA,
- title="Niche cluster")
+spatPlot2D(giotto_obj,
+ spat_unit = "cell",
+ group_by = 'list_ID',
+ cell_color = "NicheCluster",
+ color_as_factor = T,
+ point_size = 1,
+ point_border_stroke = NA,
+ title="Niche cluster")
16.1.3.5 NT score
-spatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'rna',
- group_by = 'list_ID',
- cell_color = "NTScore",
- color_as_factor = FALSE,
- cell_color_gradient = "turbo",
- point_size = 1,
- point_border_stroke = NA
- )
+spatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ group_by = 'list_ID',
+ cell_color = "NTScore",
+ color_as_factor = FALSE,
+ cell_color_gradient = "turbo",
+ point_size = 1,
+ point_border_stroke = NA
+ )
-
+
diff --git a/interoperability-with-other-frameworks.html b/interoperability-with-other-frameworks.html
index 7050533..347d3af 100644
--- a/interoperability-with-other-frameworks.html
+++ b/interoperability-with-other-frameworks.html
@@ -521,10 +521,10 @@ 15 Interoperability with other fr
15.1 Load Giotto object
To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process:
-
+
Load a Giotto mini Visium object, which will be used for demonstrating interoperability.
-
+
15.2 Seurat
@@ -533,99 +533,99 @@ 15.2 Seurat
15.2.1 Conversion of Giotto Object to Seurat Object
To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function
-
+
Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data.
-gToS <- giottoToSeuratV5(gobject = gobject,
- spat_unit = "cell")
-
-plot1 <- VlnPlot(gToS,
- features = "nCount_rna",
- pt.size = 0.1) +
- NoLegend()
-
-plot2 <- SpatialFeaturePlot(gToS,
- features = "nCount_rna",
- pt.size.factor = 2) +
- theme(legend.position = "right")
-
-wrap_plots(plot1, plot2)
+gToS <- giottoToSeuratV5(gobject = gobject,
+ spat_unit = "cell")
+
+plot1 <- VlnPlot(gToS,
+ features = "nCount_rna",
+ pt.size = 0.1) +
+ NoLegend()
+
+plot2 <- SpatialFeaturePlot(gToS,
+ features = "nCount_rna",
+ pt.size.factor = 2) +
+ theme(legend.position = "right")
+
+wrap_plots(plot1, plot2)
15.2.1.1 Apply SCTransform
We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function.
-
+
15.2.1.2 Dimension Reduction
We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object:
-
+
15.2.2 Conversion of Seurat object Back to Giotto Object
To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks.
-giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
- spatial_assay = "rna",
- dim_reduction = c("pca", "umap"),
- sp_network = "Delaunay_network",
- nn_network = c("sNN.pca", "custom_NN" ))
+giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
+ spatial_assay = "rna",
+ dim_reduction = c("pca", "umap"),
+ sp_network = "Delaunay_network",
+ nn_network = c("sNN.pca", "custom_NN" ))
15.2.2.1 Clustering and Plotting UMAP
Here we perform K-means clustering on the UMAP results obtained from the Seurat object:
-## k-means clustering
-giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
- dim_reduction_to_use = "pca")
-
-#Plot UMAP post-clustering to visualize kmeans
-graph2 <- Giotto::plotUMAP(
- gobject = giottoFromSeurat,
- cell_color = "kmeans",
- show_NN_network = TRUE,
- point_size = 2.5
-)
+## k-means clustering
+giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
+ dim_reduction_to_use = "pca")
+
+#Plot UMAP post-clustering to visualize kmeans
+graph2 <- Giotto::plotUMAP(
+ gobject = giottoFromSeurat,
+ cell_color = "kmeans",
+ show_NN_network = TRUE,
+ point_size = 2.5
+)
15.2.2.2 Spatial CoExpression
We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function:
-ranktest <- binSpect(giottoFromSeurat,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "Delaunay_network")
-
-ext_spatial_genes <- ranktest[1:300,]$feats
-
-spat_cor_netw_DT <- detectSpatialCorFeats(
- giottoFromSeurat,
- method = "network",
- spatial_network_name = "Delaunay_network",
- subset_feats = ext_spatial_genes)
-
-top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
- feats = "Dsp",
- show_top_feats = 10)
-
-spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
- name = "spat_netw_clus",
- k = 7)
-heatmSpatialCorFeats(
- giottoFromSeurat,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL),
- save_plot = TRUE,
- show_plot = TRUE,
- return_plot = FALSE,
- save_param = list(base_height = 6, base_width = 8, units = 'cm'))
+ranktest <- binSpect(giottoFromSeurat,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "Delaunay_network")
+
+ext_spatial_genes <- ranktest[1:300,]$feats
+
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ giottoFromSeurat,
+ method = "network",
+ spatial_network_name = "Delaunay_network",
+ subset_feats = ext_spatial_genes)
+
+top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
+ feats = "Dsp",
+ show_top_feats = 10)
+
+spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
+ name = "spat_netw_clus",
+ k = 7)
+heatmSpatialCorFeats(
+ giottoFromSeurat,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ return_plot = FALSE,
+ save_param = list(base_height = 6, base_width = 8, units = 'cm'))
@@ -635,60 +635,60 @@ 15.3 SpatialExperiment
To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries.
-library(SpatialExperiment)
-library(ggspavis)
-library(pheatmap)
-library(scater)
-library(scran)
-library(nnSVG)
+library(SpatialExperiment)
+library(ggspavis)
+library(pheatmap)
+library(scater)
+library(scran)
+library(nnSVG)
15.3.1 Convert Giotto Object to SpatialExperiment Object
To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function.
-
+
The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use:
-
+
15.3.1.1 Identify top spatially variable genes with nnSVG
We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate:
-# One of the assays should be "logcounts"
-# We rename the normalized assay to "logcounts"
-assayNames(spe)[[2]] <- "logcounts"
-
-# Create model matrix for leiden clustering labels
-X <- model.matrix(~ colData(spe)$leiden_clus)
-dim(X)
+# One of the assays should be "logcounts"
+# We rename the normalized assay to "logcounts"
+assayNames(spe)[[2]] <- "logcounts"
+
+# Create model matrix for leiden clustering labels
+X <- model.matrix(~ colData(spe)$leiden_clus)
+dim(X)
- Run nnSVG
This step will take several minutes to run
-
+
15.3.2 Conversion of SpatialExperiment object back to Giotto
We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object.
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-print(giottoFromSPE)
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+print(giottoFromSPE)
15.3.2.1 Plotting top genes from nnSVG in Giotto
Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object.
-ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
- output = "data.table")
-
-ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
-spatFeatPlot2D(giottoFromSPE,
- expression_values = "scaled_rna_cell",
- feats = ext_spatial_genes[1:4],
- point_size = 2)
+ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
+ output = "data.table")
+
+ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
+spatFeatPlot2D(giottoFromSPE,
+ expression_values = "scaled_rna_cell",
+ feats = ext_spatial_genes[1:4],
+ point_size = 2)
@@ -701,97 +701,97 @@ 15.4 AnnData
15.4.1 Load Required Libraries
To start, we need to load the necessary libraries, including reticulate for interfacing with Python.
-
+
15.4.2 Specify Path for Results
First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions.
-# Specify path to which results may be saved
-results_directory <- "results/03_session4/giotto_anndata_conversion/"
-
-instrs <- showGiottoInstructions(gobject)
-
-mini_gobject <- replaceGiottoInstructions(gobject = gobject,
- instructions = instrs)
+# Specify path to which results may be saved
+results_directory <- "results/03_session4/giotto_anndata_conversion/"
+
+instrs <- showGiottoInstructions(gobject)
+
+mini_gobject <- replaceGiottoInstructions(gobject = gobject,
+ instructions = instrs)
15.4.2.1 Create Default kNN Network
We will create a k-nearest neighbor (kNN) network using mostly default parameters.
-
+
15.4.3 Giotto To AnnData
To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData()
-
+
Next, we import scanpy and perform a series of preprocessing steps on the AnnData object.
-scanpy <- import("scanpy")
-
-adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
-
-# Normalize total counts per cell
-scanpy$pp$normalize_total(adata, target_sum=1e4)
-
-# Log-transform the data
-scanpy$pp$log1p(adata)
-
-# Perform PCA
-scanpy$pp$pca(adata, n_comps=40L)
-
-# Compute the neighborhood graph
-scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
-
-# Run UMAP
-scanpy$tl$umap(adata)
-
-# Save the processed AnnData object
-adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
-
-processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
+scanpy <- import("scanpy")
+
+adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
+
+# Normalize total counts per cell
+scanpy$pp$normalize_total(adata, target_sum=1e4)
+
+# Log-transform the data
+scanpy$pp$log1p(adata)
+
+# Perform PCA
+scanpy$pp$pca(adata, n_comps=40L)
+
+# Compute the neighborhood graph
+scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
+
+# Run UMAP
+scanpy$tl$umap(adata)
+
+# Save the processed AnnData object
+adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
+
+processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
15.4.4 Convert AnnData to Giotto
Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto.
-
+
15.4.4.1 UMAP Visualization
Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object.
-Giotto::plotUMAP(
- gobject = giottoFromAnndata,
- dim_reduction_name = "umap.ad",
- cell_color = "leiden_clus",
- point_size = 3
-)
+Giotto::plotUMAP(
+ gobject = giottoFromAnndata,
+ dim_reduction_name = "umap.ad",
+ cell_color = "leiden_clus",
+ point_size = 3
+)
15.5 Create mini Vizgen object
-mini_gobject <- loadGiottoMini(dataset = "vizgen",
- python_path = NULL)
-
-mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
- instructions = instrs)
-mini_gobject <- createNearestNetwork(gobject = mini_gobject,
- spat_unit = "aggregate",
- feat_type = "rna",
- type = "kNN",
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap",
- k = 6,
- name = "new_network")
+mini_gobject <- loadGiottoMini(dataset = "vizgen",
+ python_path = NULL)
+
+mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
+ instructions = instrs)
+mini_gobject <- createNearestNetwork(gobject = mini_gobject,
+ spat_unit = "aggregate",
+ feat_type = "rna",
+ type = "kNN",
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap",
+ k = 6,
+ name = "new_network")
Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details.
-
+
diff --git a/introduction-to-the-giotto-package.html b/introduction-to-the-giotto-package.html
index e22d6d1..0ab2014 100644
--- a/introduction-to-the-giotto-package.html
+++ b/introduction-to-the-giotto-package.html
@@ -546,19 +546,58 @@ 4.3 Installation + python environ
4.3.1 Giotto installation
Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository.
Much of this already covered in Section 2.2, but the highlights are:
-
-- Ensure your system
-
-To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak.
+
+4.3.1.1 System prerequisites
+
+- for windows, Rtools needs to be installed
+- a major dependency terra needs
GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux
+
+
+
+4.3.1.2 Installation of released version
+To install the currently released version of Giotto in a single step:
+This should automatically install all the Giotto dependencies and other Giotto module packages.
+
+
+4.3.1.3 Installation of dev branch Giotto packages
+You can install dev branch versions by using devtools::install_github()
+Core module dev branchs:
+
+- “drieslab/Giotto@suite_dev”
+- “drieslab/GiottoVisuals@dev”
+- “drieslab/GiottoClass@dev”
+- “drieslab/GiottoUtils@dev”
+
+
+pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages.
+
+
+4.3.1.4 Common install issues
+If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work.
+
+If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it.
+
+## # Downloading packages -------------------------------------------------------
+## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s]
+## Successfully downloaded 1 package in 1.2 seconds.
+##
+## The following package(s) will be installed:
+## - irlba [2.3.5.1]
+## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu".
+##
+## # Installing packages --------------------------------------------------------
+## - Installing irlba ... OK [built from source and cached in 5.7s]
+## Successfully installed 1 package in 5.7 seconds.
+
4.3.2 Python environment
4.3.2.1 Default installation
In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment.
-
+
Two things are needed for python to work:
- A conda (e.g. miniconda or anaconda) installation which is the package and environment management system.
@@ -580,17 +619,17 @@ 4.3.2.1 Default installation
4.3.2.2 Custom installs
Custom python environments can be made by first setting up a new environment and establishing the name and python version to use.
-
+
Following that, one or more python packages to install can be added to the environment.
-reticulate::py_install(
- pip = TRUE,
- envname = '[name of env]',
- packages = c(
- "package1",
- "package2",
- "..."
- )
-)
+reticulate::py_install(
+ pip = TRUE,
+ envname = '[name of env]',
+ packages = c(
+ "package1",
+ "package2",
+ "..."
+ )
+)
Once an environment has been set up, Giotto can hook into it.
@@ -612,17 +651,56 @@ 4.3.2.3 Using a specific environm
Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto.
You would run file.edit("~/.Rprofile") and then add options("giotto.py_path" = "[path to env or envname]") as a line so that it is automatically set at the start of each session.
If a specific environment should only be used a couple times then method 1 is easiest:
-
+
To check which conda environments exist on your machine:
-
+
Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running:
-
+
4.4 Giotto instructions
-Giotto
+Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are:
+
+- python_path - when set, will activate a python environment
+- save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render.
+- save_plot - whether to save plots to the
save_dir
+- return_plot - whether to return the plot objects. When FALSE, only NULL is returned
+- show_plot - whether to show the plot in the viewer
+
+These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function.
+library(Giotto)
+save_dir <- "results/01_session2/"
+
+# this call will also intialize the python env
+instrs <- createGiottoInstructions(
+ save_dir = save_dir, # working directory is the default
+ show_plot = FALSE,
+ save_plot = TRUE,
+ return_plot = FALSE,
+ python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default
+)
+force(instrs)
+Giotto object creation functions all have an instructions param for passing in instructions objects.
+giotto objects will also respond to the instructions() generic.
+test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object
+
+# example plot
+g <- GiottoData::loadGiottoMini("visium")
+instructions(g) <- instrs
+instructions(g, "show_plot") # instructions say not to plot to viewer
+spatPlot2D(g, show_image = TRUE, image_name = "image")
+# instead it will directly write to the results folder
+As an example, you can also set individual instructions
+
+
+
+
+Figure 4.2: example image output
+
+
diff --git a/multi-omics-integration.html b/multi-omics-integration.html
index 47cad23..e76e25a 100644
--- a/multi-omics-integration.html
+++ b/multi-omics-integration.html
@@ -520,14 +520,14 @@ 14 Multi-omics integration
14.1 The CytAssist technology
The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image.
-
+
Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics.
The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument.
-
+
Figure 14.2: CytAssist workflow. Source: 10X genomics.
@@ -542,19 +542,19 @@
14.2 Introduction to the spatial
14.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/03_session3")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
+dir.create("data/03_session3")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”.
-
+
14.4 Create the Giotto object
@@ -564,124 +564,124 @@ 14.4 Create the Giotto objectx,y(,z) coordinates for cells or spots (or the path to)
createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/03_session3/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-# Provide the path to the data folder
-data_path <- "data/03_session3/"
-
-# Create object directly from the data folder
-visium_tonsil <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/03_session3/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+# Provide the path to the data folder
+data_path <- "data/03_session3/"
+
+# Create object directly from the data folder
+visium_tonsil <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
- Print the information of the object, note that both rna and protein are listed in the expression slot.
-
+
14.5 Subset on spots that were covered by tissue
-spatPlot2D(
- gobject = visium_tonsil,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE,
- image_name = "image"
-)
-
+spatPlot2D(
+ gobject = visium_tonsil,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE,
+ image_name = "image"
+)
+
Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object.
-
+
14.6 RNA processing
- Run the Filtering, normalization, and statistics steps using only the RNA feature.
-visium_tonsil <- filterGiotto(
- gobject = visium_tonsil,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil)
+visium_tonsil <- filterGiotto(
+ gobject = visium_tonsil,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1000,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil)
- Dimension reduction
Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs.
-visium_tonsil <- calculateHVF(gobject = visium_tonsil)
-
-visium_tonsil <- runPCA(gobject = visium_tonsil)
+visium_tonsil <- calculateHVF(gobject = visium_tonsil)
+
+visium_tonsil <- runPCA(gobject = visium_tonsil)
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the RNA-based Leiden clusters.
-
+
- Visualization
Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot.
-plotUMAP(gobject = visium_tonsil,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters.
Plot the spatial distribution of the RNA-based Leiden cluster per spot.
-spatPlot2D(gobject = visium_tonsil,
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.5: Spatial distribution of RNA-based Leiden clusters.
@@ -693,80 +693,80 @@
14.7 Protein processingvisium_tonsil <- filterGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein")
+visium_tonsil <- filterGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein")
- Dimension reduction
Calculate the principal components using all the proteins available in the dataset.
-
+
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the Protein-based Leiden clusters.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- resolution = 1,
- n_iterations = 1000)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ resolution = 1,
+ n_iterations = 1000)
- Visualization
Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters.
Plot the spatial distribution of the Protein-based Leiden clusters.
-spatPlot2D(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.7: Spatial distribution of Protein-based Leiden clusters.
@@ -779,68 +779,68 @@
14.8 Multi-omics integrationCalculate kNN
Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type.
-## RNA modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
-
-## Protein modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
+## RNA modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
+
+## Protein modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
- Run WNN
Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object.
-visium_tonsil <- runWNN(visium_tonsil,
- spat_unit = "cell",
- modality_1 = "rna",
- modality_2 = "protein",
- pca_name_modality_1 = "pca",
- pca_name_modality_2 = "protein.pca",
- k = 20,
- integrated_feat_type = NULL,
- matrix_result_name = NULL,
- w_name_modality_1 = NULL,
- w_name_modality_2 = NULL,
- verbose = TRUE)
+visium_tonsil <- runWNN(visium_tonsil,
+ spat_unit = "cell",
+ modality_1 = "rna",
+ modality_2 = "protein",
+ pca_name_modality_1 = "pca",
+ pca_name_modality_2 = "protein.pca",
+ k = 20,
+ integrated_feat_type = NULL,
+ matrix_result_name = NULL,
+ w_name_modality_1 = NULL,
+ w_name_modality_2 = NULL,
+ verbose = TRUE)
- Run Integrated umap
Calculate the UMAP using the weights of each feature per spot.
-visium_tonsil <- runIntegratedUMAP(visium_tonsil,
- modality1 = "rna",
- modality2 = "protein",
- spread = 7,
- min_dist = 1,
- force = FALSE)
+visium_tonsil <- runIntegratedUMAP(visium_tonsil,
+ modality1 = "rna",
+ modality2 = "protein",
+ spread = 7,
+ min_dist = 1,
+ force = FALSE)
- Calculate integrated clusters
Calculate the multiomics-based Leiden clusters using the weights of each feature per spot.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- nn_network_to_use = "kNN",
- network_name = "integrated_kNN",
- name = "integrated_leiden_clus",
- resolution = 1)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ nn_network_to_use = "kNN",
+ network_name = "integrated_kNN",
+ name = "integrated_leiden_clus",
+ resolution = 1)
- Visualize the integrated UMAP
Plot the integrated UMAP and color the spots using the integrated Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- dim_reduction_name = "integrated.umap",
- point_size = 2,
- title = "Integrated UMAP using Integrated Leiden clusters")
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ dim_reduction_name = "integrated.umap",
+ point_size = 2,
+ title = "Integrated UMAP using Integrated Leiden clusters")
+
Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters.
@@ -850,14 +850,14 @@
14.8 Multi-omics integrationVisualize spatial plot with integrated clusters
Plot the spatial distribution of the integrated Leiden clusters.
-spatPlot2D(visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- point_size = 3,
- show_image = TRUE,
- title = "Integrated Leiden clustering")
-
+spatPlot2D(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ point_size = 3,
+ show_image = TRUE,
+ title = "Integrated Leiden clustering")
+
Figure 14.9: Spatial distribution of the integrated Leiden clusters.
@@ -866,85 +866,85 @@
14.8 Multi-omics integration
14.9 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
- [7] matrixStats_1.3.0 compiler_4.4.1
- [9] png_0.1-8 systemfonts_1.1.0
- [11] vctrs_0.6.5 reshape2_1.4.4
- [13] stringr_1.5.1 pkgconfig_2.0.3
- [15] SpatialExperiment_1.14.0 crayon_1.5.3
- [17] fastmap_1.2.0 backports_1.5.0
- [19] magick_2.8.4 XVector_0.44.0
- [21] labeling_0.4.3 utf8_1.2.4
- [23] rmarkdown_2.27 UCSC.utils_1.0.0
- [25] ragg_1.3.2 purrr_1.0.2
- [27] xfun_0.46 beachmat_2.20.0
- [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
- [31] jsonlite_1.8.8 DelayedArray_0.30.1
- [33] BiocParallel_1.38.0 terra_1.7-78
- [35] irlba_2.3.5.1 parallel_4.4.1
- [37] R6_2.5.1 stringi_1.8.4
- [39] RColorBrewer_1.1-3 reticulate_1.38.0
- [41] parallelly_1.37.1 GenomicRanges_1.56.1
- [43] scattermore_1.2 Rcpp_1.0.13
- [45] bookdown_0.40 SummarizedExperiment_1.34.0
- [47] knitr_1.48 future.apply_1.11.2
- [49] R.utils_2.12.3 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
+ [7] matrixStats_1.3.0 compiler_4.4.1
+ [9] png_0.1-8 systemfonts_1.1.0
+ [11] vctrs_0.6.5 reshape2_1.4.4
+ [13] stringr_1.5.1 pkgconfig_2.0.3
+ [15] SpatialExperiment_1.14.0 crayon_1.5.3
+ [17] fastmap_1.2.0 backports_1.5.0
+ [19] magick_2.8.4 XVector_0.44.0
+ [21] labeling_0.4.3 utf8_1.2.4
+ [23] rmarkdown_2.27 UCSC.utils_1.0.0
+ [25] ragg_1.3.2 purrr_1.0.2
+ [27] xfun_0.46 beachmat_2.20.0
+ [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
+ [31] jsonlite_1.8.8 DelayedArray_0.30.1
+ [33] BiocParallel_1.38.0 terra_1.7-78
+ [35] irlba_2.3.5.1 parallel_4.4.1
+ [37] R6_2.5.1 stringi_1.8.4
+ [39] RColorBrewer_1.1-3 reticulate_1.38.0
+ [41] parallelly_1.37.1 GenomicRanges_1.56.1
+ [43] scattermore_1.2 Rcpp_1.0.13
+ [45] bookdown_0.40 SummarizedExperiment_1.34.0
+ [47] knitr_1.48 future.apply_1.11.2
+ [49] R.utils_2.12.3 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/reference-keys.txt b/reference-keys.txt
index a17604d..555f956 100644
--- a/reference-keys.txt
+++ b/reference-keys.txt
@@ -4,115 +4,116 @@ fig:unnamed-chunk-14
fig:unnamed-chunk-15
fig:unnamed-chunk-16
fig:unnamed-chunk-18
-fig:unnamed-chunk-30
-fig:unnamed-chunk-35
-fig:unnamed-chunk-38
-fig:unnamed-chunk-40
+fig:unnamed-chunk-32
+fig:unnamed-chunk-37
fig:unnamed-chunk-42
-fig:unnamed-chunk-44
+fig:unnamed-chunk-45
+fig:unnamed-chunk-47
fig:unnamed-chunk-49
fig:unnamed-chunk-51
-fig:unnamed-chunk-53
-fig:unnamed-chunk-55
-fig:unnamed-chunk-59
-fig:unnamed-chunk-61
-fig:unnamed-chunk-63
+fig:unnamed-chunk-56
+fig:unnamed-chunk-58
+fig:unnamed-chunk-60
+fig:unnamed-chunk-62
fig:unnamed-chunk-66
-fig:unnamed-chunk-69
-fig:unnamed-chunk-74
+fig:unnamed-chunk-68
+fig:unnamed-chunk-70
+fig:unnamed-chunk-73
fig:unnamed-chunk-76
-fig:unnamed-chunk-78
-fig:unnamed-chunk-80
-fig:unnamed-chunk-82
-fig:unnamed-chunk-84
+fig:unnamed-chunk-81
+fig:unnamed-chunk-83
+fig:unnamed-chunk-85
fig:unnamed-chunk-87
+fig:unnamed-chunk-89
+fig:unnamed-chunk-91
fig:unnamed-chunk-94
-fig:unnamed-chunk-96
-fig:unnamed-chunk-98
fig:unnamed-chunk-101
fig:unnamed-chunk-103
fig:unnamed-chunk-105
+fig:unnamed-chunk-108
+fig:unnamed-chunk-110
fig:unnamed-chunk-112
-fig:unnamed-chunk-114
-fig:unnamed-chunk-118
-fig:unnamed-chunk-120
-fig:unnamed-chunk-123
-fig:unnamed-chunk-128
-fig:unnamed-chunk-131
-fig:unnamed-chunk-134
-fig:unnamed-chunk-137
+fig:unnamed-chunk-119
+fig:unnamed-chunk-121
+fig:unnamed-chunk-125
+fig:unnamed-chunk-127
+fig:unnamed-chunk-130
+fig:unnamed-chunk-135
+fig:unnamed-chunk-138
fig:unnamed-chunk-141
-fig:unnamed-chunk-143
-fig:unnamed-chunk-147
+fig:unnamed-chunk-144
+fig:unnamed-chunk-148
fig:unnamed-chunk-150
-fig:unnamed-chunk-152
+fig:unnamed-chunk-154
+fig:unnamed-chunk-157
fig:unnamed-chunk-159
-fig:unnamed-chunk-161
-fig:unnamed-chunk-163
-fig:unnamed-chunk-165
-fig:unnamed-chunk-167
-fig:unnamed-chunk-169
-fig:unnamed-chunk-171
+fig:unnamed-chunk-166
+fig:unnamed-chunk-168
+fig:unnamed-chunk-170
+fig:unnamed-chunk-172
fig:unnamed-chunk-174
-fig:unnamed-chunk-175
fig:unnamed-chunk-176
-fig:unnamed-chunk-185
-fig:unnamed-chunk-191
-fig:unnamed-chunk-194
+fig:unnamed-chunk-178
+fig:unnamed-chunk-181
+fig:unnamed-chunk-182
+fig:unnamed-chunk-183
+fig:unnamed-chunk-192
+fig:unnamed-chunk-198
fig:unnamed-chunk-201
-fig:unnamed-chunk-204
-fig:unnamed-chunk-206
-fig:unnamed-chunk-210
-fig:unnamed-chunk-212
-fig:unnamed-chunk-215
+fig:unnamed-chunk-208
+fig:unnamed-chunk-211
+fig:unnamed-chunk-213
fig:unnamed-chunk-217
fig:unnamed-chunk-219
-fig:unnamed-chunk-221
+fig:unnamed-chunk-222
fig:unnamed-chunk-224
-fig:unnamed-chunk-225
-fig:unnamed-chunk-227
-fig:unnamed-chunk-229
-fig:unnamed-chunk-230
+fig:unnamed-chunk-226
+fig:unnamed-chunk-228
+fig:unnamed-chunk-231
fig:unnamed-chunk-232
fig:unnamed-chunk-234
fig:unnamed-chunk-236
-fig:unnamed-chunk-275
-fig:unnamed-chunk-279
-fig:unnamed-chunk-281
-fig:unnamed-chunk-283
+fig:unnamed-chunk-237
+fig:unnamed-chunk-239
+fig:unnamed-chunk-241
+fig:unnamed-chunk-243
+fig:unnamed-chunk-282
fig:unnamed-chunk-286
fig:unnamed-chunk-288
fig:unnamed-chunk-290
-fig:unnamed-chunk-292
-fig:unnamed-chunk-294
+fig:unnamed-chunk-293
+fig:unnamed-chunk-295
fig:unnamed-chunk-297
fig:unnamed-chunk-299
fig:unnamed-chunk-301
-fig:unnamed-chunk-303
-fig:unnamed-chunk-305
-fig:unnamed-chunk-365
-fig:unnamed-chunk-366
+fig:unnamed-chunk-304
+fig:unnamed-chunk-306
+fig:unnamed-chunk-308
+fig:unnamed-chunk-310
+fig:unnamed-chunk-312
fig:unnamed-chunk-372
+fig:unnamed-chunk-373
fig:unnamed-chunk-379
-fig:unnamed-chunk-381
-fig:unnamed-chunk-387
-fig:unnamed-chunk-389
-fig:unnamed-chunk-395
-fig:unnamed-chunk-397
-fig:unnamed-chunk-462
-fig:unnamed-chunk-465
+fig:unnamed-chunk-386
+fig:unnamed-chunk-388
+fig:unnamed-chunk-394
+fig:unnamed-chunk-396
+fig:unnamed-chunk-402
+fig:unnamed-chunk-404
fig:unnamed-chunk-469
-fig:unnamed-chunk-470
fig:unnamed-chunk-472
-fig:unnamed-chunk-474
fig:unnamed-chunk-476
-fig:unnamed-chunk-478
+fig:unnamed-chunk-477
fig:unnamed-chunk-479
fig:unnamed-chunk-481
+fig:unnamed-chunk-483
fig:unnamed-chunk-485
+fig:unnamed-chunk-486
fig:unnamed-chunk-488
-fig:unnamed-chunk-490
fig:unnamed-chunk-492
+fig:unnamed-chunk-495
+fig:unnamed-chunk-497
+fig:unnamed-chunk-499
giotto-suite-workshop-2024
instructors
topics-and-schedule
@@ -153,6 +154,10 @@ presentation-1
ecosystem
installation-python-environment
giotto-installation-1
+system-prerequisites
+installation-of-released-version
+installation-of-dev-branch-giotto-packages
+common-install-issues
python-environment
default-installation
custom-installs
diff --git a/search_index.json b/search_index.json
index 2736898..72ab7c3 100644
--- a/search_index.json
+++ b/search_index.json
@@ -1 +1 @@
-[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: Ensure your system To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak. pak::pak("drieslab/Giotto") 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
+[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: 4.3.1.1 System prerequisites for windows, Rtools needs to be installed a major dependency terra needs GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux 4.3.1.2 Installation of released version To install the currently released version of Giotto in a single step: pak::pak("drieslab/Giotto") This should automatically install all the Giotto dependencies and other Giotto module packages. 4.3.1.3 Installation of dev branch Giotto packages You can install dev branch versions by using devtools::install_github() Core module dev branchs: “drieslab/Giotto@suite_dev” “drieslab/GiottoVisuals@dev” “drieslab/GiottoClass@dev” “drieslab/GiottoUtils@dev” devtools::install_github("drieslab/GiottoClass@dev") pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages. 4.3.1.4 Common install issues If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work. devtools::install_version("Matrix", version = "1.6-5") If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it. install.packages("irlba", type = "source") ## # Downloading packages ------------------------------------------------------- ## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s] ## Successfully downloaded 1 package in 1.2 seconds. ## ## The following package(s) will be installed: ## - irlba [2.3.5.1] ## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu". ## ## # Installing packages -------------------------------------------------------- ## - Installing irlba ... OK [built from source and cached in 5.7s] ## Successfully installed 1 package in 5.7 seconds. 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are: python_path - when set, will activate a python environment save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render. save_plot - whether to save plots to the save_dir return_plot - whether to return the plot objects. When FALSE, only NULL is returned show_plot - whether to show the plot in the viewer These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function. library(Giotto) save_dir <- "results/01_session2/" # this call will also intialize the python env instrs <- createGiottoInstructions( save_dir = save_dir, # working directory is the default show_plot = FALSE, save_plot = TRUE, return_plot = FALSE, python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default ) force(instrs) Giotto object creation functions all have an instructions param for passing in instructions objects. giotto objects will also respond to the instructions() generic. test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object # example plot g <- GiottoData::loadGiottoMini("visium") instructions(g) <- instrs instructions(g, "show_plot") # instructions say not to plot to viewer spatPlot2D(g, show_image = TRUE, image_name = "image") # instead it will directly write to the results folder As an example, you can also set individual instructions instructions(g, "show_plot") <- TRUE spatPlot2D(g, show_image = TRUE, image_name = "image") Figure 4.2: example image output "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
diff --git a/spatial-multi-modal-analysis.html b/spatial-multi-modal-analysis.html
index 5bdab74..c78f952 100644
--- a/spatial-multi-modal-analysis.html
+++ b/spatial-multi-modal-analysis.html
@@ -563,48 +563,48 @@ 13.2.3 Spatial classes:# load in data
-library(Giotto)
-
-g <- GiottoData::loadGiottoMini("vizgen")
-
-activeSpatUnit(g) <- "aggregate"
-
-gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE)
-
-gimg <- getGiottoImage(g)
+
13.3 Examples of the simple transforms with a giottoPolygon
-rain <- rainbow(nrow(gpoly))
-
-line_width <- 0.3
-
-# par to setup the grid plotting layout
-p <- par(no.readonly = TRUE)
-par(mfrow=c(3,3))
-gpoly |>
- plot(main = "no transform", col = rain, lwd = line_width)
-
-gpoly |> spatShift(dx = 1000) |>
- plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
-
-gpoly |> spin(45) |>
- plot(main = "spin(45)", col = rain, lwd = line_width)
-
-gpoly |> rescale(fx = 10, fy = 6) |>
- plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
-
-gpoly |> flip(direction = "vertical") |>
- plot(main = "flip()", col = rain, lwd = line_width)
-
-gpoly |> t() |>
- plot(main = "t()", col = rain, lwd = line_width)
-
-gpoly |> shear(fx = 0.5) |>
- plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
-par(p)
+rain <- rainbow(nrow(gpoly))
+
+line_width <- 0.3
+
+# par to setup the grid plotting layout
+p <- par(no.readonly = TRUE)
+par(mfrow=c(3,3))
+gpoly |>
+ plot(main = "no transform", col = rain, lwd = line_width)
+
+gpoly |> spatShift(dx = 1000) |>
+ plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
+
+gpoly |> spin(45) |>
+ plot(main = "spin(45)", col = rain, lwd = line_width)
+
+gpoly |> rescale(fx = 10, fy = 6) |>
+ plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
+
+gpoly |> flip(direction = "vertical") |>
+ plot(main = "flip()", col = rain, lwd = line_width)
+
+gpoly |> t() |>
+ plot(main = "t()", col = rain, lwd = line_width)
+
+gpoly |> shear(fx = 0.5) |>
+ plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
+par(p)
@@ -617,20 +617,20 @@ 13.4 Affine transforms# create affine2d
-aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
+# create affine2d
+aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set.
-
-# append several simple transforms
-aff <- aff |>
- spatShift(dx = 1000) |>
- spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- flip(direction = "vertical") |>
- t() |>
- shear(fx = 0.5)
-force(aff)
+
+# append several simple transforms
+aff <- aff |>
+ spatShift(dx = 1000) |>
+ spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ flip(direction = "vertical") |>
+ t() |>
+ shear(fx = 0.5)
+force(aff)
<affine2d>
anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax)
rotate : -0.785398163397448 (rad)
@@ -639,35 +639,35 @@ 13.4 Affine transformsplot(aff)
+
We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests.
-
+
13.5 Image transforms
Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting.
spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image.
-# affine transform of image info matches with polygon info
-gimg |> affine(aff) |> plot()
-gpoly |> affine(aff) |>
- plot(add = TRUE,
- border = "cyan",
- lwd = 0.3)
+# affine transform of image info matches with polygon info
+gimg |> affine(aff) |> plot()
+gpoly |> affine(aff) |>
+ plot(add = TRUE,
+ border = "cyan",
+ lwd = 0.3)
-# affine of the giotto object
-g |> affine(aff) |>
- spatInSituPlotPoints(
- show_image = TRUE,
- feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
- feats_color_code = rainbow(4),
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- point_size = 0.1,
- use_overlap = FALSE
- )
+# affine of the giotto object
+g |> affine(aff) |>
+ spatInSituPlotPoints(
+ show_image = TRUE,
+ feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
+ feats_color_code = rainbow(4),
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ point_size = 0.1,
+ use_overlap = FALSE
+ )
Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used.
@@ -687,256 +687,256 @@ 13.6.1 Example dataset: Xenium Br
– cooresponding centroid locations
The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance.
-library(Giotto)
-instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
- save_plot = TRUE,
- show_plot = TRUE)
-options(timeout = 999999)
-download_dir <-file.path(getwd(),'/data/03_session2/')
-destfile <- file.path(download_dir,'Multimodal_registration.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
-Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
-Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
+library(Giotto)
+instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
+ save_plot = TRUE,
+ show_plot = TRUE)
+options(timeout = 999999)
+download_dir <-file.path(getwd(),'/data/03_session2/')
+destfile <- file.path(download_dir,'Multimodal_registration.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
+Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
+Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
13.6.2 Target Coordinate system
Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system.
-xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
-xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
-xen_cell_pl
+xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
+xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+xen_cell_pl
13.6.3 Visium to register
Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration
-G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
- gene_column_index = 2,
- png_name = 'tissue_hires_image.png',
- instructions = NULL)
-# In the meantime, calculate statistics for easier plot showing
-G_visium <- normalizeGiotto(G_visium)
-G_visium <- addStatistics(G_visium)
-V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
-V_origin
+G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
+ gene_column_index = 2,
+ png_name = 'tissue_hires_image.png',
+ instructions = NULL)
+# In the meantime, calculate statistics for easier plot showing
+G_visium <- normalizeGiotto(G_visium)
+G_visium <- addStatistics(G_visium)
+V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
+V_origin
The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform.
-# create affine2d
-aff <- affine(diag(c(1,1)))
-aff <- aff |>
- spin(90) |>
- flip(direction = "horizontal")
-force(aff)
-
-# Apply the transform
-V_tansformed <- affine(G_visium,aff)
-spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
-spatplot_to_register
+# create affine2d
+aff <- affine(diag(c(1,1)))
+aff <- aff |>
+ spin(90) |>
+ flip(direction = "horizontal")
+force(aff)
+
+# Apply the transform
+V_tansformed <- affine(G_visium,aff)
+spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
+spatplot_to_register
Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images.
Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app.
-
+
Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result.
-landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
-affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
-
-V_final <- affine(G_visium,affine_mtx %*% aff@affine)
-
-spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
-spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
+affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
+
+V_final <- affine(G_visium,affine_mtx %*% aff@affine)
+
+spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
+spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.3.1 Create Pseudo Visium dataset for comparison
Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance.
-V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
-spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
-spatPlot2D(V_final,
- show_network = T,
- network_color = 'blue',
- point_size = 1)
-center_to_center <- min(spat_network$distance)
-radius <- center_to_center*55/200
+V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
+spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
+spatPlot2D(V_final,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1)
+center_to_center <- min(spat_network$distance)
+radius <- center_to_center*55/200
Now we get the Pseudo Visium polygons
-Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
-stamp_dt <- circleVertices(radius = radius, npoints = 100)
-pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
-pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
-plot(pseudo_visium_poly)
+Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
+stamp_dt <- circleVertices(radius = radius, npoints = 100)
+pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
+pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
+plot(pseudo_visium_poly)
Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points
-xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
-gpoints <- createGiottoPoints(xen_transcripts)
-Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
- gpolygons = list('visium' = pseudo_visium_poly))
+xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
+gpoints <- createGiottoPoints(xen_transcripts)
+Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
+ gpolygons = list('visium' = pseudo_visium_poly))
Get gene expression information by overlapping polygon to points
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'visium')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "visium",
- feat_info = "rna",
- aggr_function = "sum")
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'visium')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "visium",
+ feat_info = "rna",
+ aggr_function = "sum")
Manipulate the expression for plotting
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'visium',
- name = 'plot')
-
-spatFeatPlot2D(Xen_obj,
- point_size = 3.5,
- expression_values = 'plot',
- show_image = F,
- feats = 'ERBB2')
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ name = 'plot')
+
+spatFeatPlot2D(Xen_obj,
+ point_size = 3.5,
+ expression_values = 'plot',
+ show_image = F,
+ feats = 'ERBB2')
Subset the registered Visium and plot same gene
-#get the extent of giotto points, xmin, xmax, ymin, ymax
-subset_extent <- ext(gpoints@spatVector)
-sub_visium <- subsetGiottoLocs(V_final,
- x_min = subset_extent[1],
- x_max = subset_extent[2],
- y_min = subset_extent[3],
- y_max = subset_extent[4])
-spatFeatPlot2D(sub_visium,
- point_size = 2,
- expression_values = 'scaled',
- show_image = F,
- feats = 'ERBB2')
+#get the extent of giotto points, xmin, xmax, ymin, ymax
+subset_extent <- ext(gpoints@spatVector)
+sub_visium <- subsetGiottoLocs(V_final,
+ x_min = subset_extent[1],
+ x_max = subset_extent[2],
+ y_min = subset_extent[3],
+ y_max = subset_extent[4])
+spatFeatPlot2D(sub_visium,
+ point_size = 2,
+ expression_values = 'scaled',
+ show_image = F,
+ feats = 'ERBB2')
13.6.4 Register post-Xenium H&E and IF image
For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics.
-HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
- imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
- micron = 0.2125)
-plot(HE_xen)
+HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
+ imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
+ micron = 0.2125)
+plot(HE_xen)
The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use.
-HE_xen <- HE_xen |> flip(direction = "vertical")
-gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
-plot(gimg_rast)
-#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
+HE_xen <- HE_xen |> flip(direction = "vertical")
+gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
+plot(gimg_rast)
+#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots
-gg <- ggplot2::ggplot()
-pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
-pl + ggplot2::geom_smooth() +
- ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+gg <- ggplot2::ggplot()
+pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
+pl + ggplot2::geom_smooth() +
+ ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.4.1 Add registered image information and compare RNA vs protein expression
With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF
-CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
-HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
-Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
- largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
+CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
+HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
+Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
+ largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
Get the cell polygons, as Xenium and IF are both subcellular resolution
-cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
-colnames(cellpoly_dt) <- c('poly_ID','x','y')
-cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
-Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
+cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
+colnames(cellpoly_dt) <- c('poly_ID','x','y')
+cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
+Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
Compute the gene expression matrix by overlay the cell polygons and giotto points.
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'cell')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "cell",
- feat_info = "rna",
- aggr_function = "sum")
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'rna',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'ERBB2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'cell')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "cell",
+ feat_info = "rna",
+ aggr_function = "sum")
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'rna',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'ERBB2',
+ point_size = 0.05)
Now we overlay the HER2 expression from the raster image with the cell polygons.
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'cell',
- image_names = c('HER2','CD20'))
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'HER2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'cell',
+ image_names = c('HER2','CD20'))
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'HER2',
+ point_size = 0.05)
We can also overlay the protein expression to Visium spots
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'visium',
- image_names = c('HER2','CD20'))
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "visium",
- feat_info = "protein",
- aggr_function = "sum")
-
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'visium',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'visium',
- feats = 'HER2',
- point_size = 2)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'visium',
+ image_names = c('HER2','CD20'))
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "visium",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'visium',
+ feats = 'HER2',
+ point_size = 2)
@@ -944,29 +944,29 @@ 13.6.4.1 Add registered image inf
13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation
Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially.
Here, we provide an example of two compressed images to show the automatic alignment pipeline.
-
+
-IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
-terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
+IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
+terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform.
-estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
- flip_horizontal = T,
- use_single_channel = T,
- single_channel_number = 3),
- y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
- invert = T),
- plot_match = T,
- max_ratio = 0.5,estimate_fun = 'Projective')
+estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
+ flip_horizontal = T,
+ use_single_channel = T,
+ single_channel_number = 3),
+ y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
+ invert = T),
+ plot_match = T,
+ max_ratio = 0.5,estimate_fun = 'Projective')
Use the estimation, we can quickly visualize the transformation
-mtx <- as.matrix(estimation$params)
-transformed <- affine(IF, mtx)
-To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
-plot(HE)
-plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
-SIFT_registration_overlay
+mtx <- as.matrix(estimation$params)
+transformed <- affine(IF, mtx)
+To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
+plot(HE)
+plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
+SIFT_registration_overlay
diff --git a/spatial-proteomics-multiplexed-immunofluorescence.html b/spatial-proteomics-multiplexed-immunofluorescence.html
index 684d4e8..f550962 100644
--- a/spatial-proteomics-multiplexed-immunofluorescence.html
+++ b/spatial-proteomics-multiplexed-immunofluorescence.html
@@ -518,33 +518,33 @@ 11 Spatial proteomics: Multiplexe
Junxiang Xu
August 6th 2024
Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so.
-#Install the Giotto Environment with Cellpose, note that we only need to do it once
-reticulate::conda_create(envname = "giotto_cellpose",
- python_version = 3.8)
-#.re.restartR()
-reticulate::use_condaenv('giotto_cellpose')
-reticulate::py_install(
- pip = TRUE,
- envname = 'giotto_cellpose',
- packages = c(
- "pandas",
- "networkx",
- "python-igraph",
- "leidenalg",
- "scikit-learn",
- "cellpose",
- "smfishhmrf",
- 'tifffile',
- 'scikit-image'
- )
-)
-#.rs.restartR()
+#Install the Giotto Environment with Cellpose, note that we only need to do it once
+reticulate::conda_create(envname = "giotto_cellpose",
+ python_version = 3.8)
+#.re.restartR()
+reticulate::use_condaenv('giotto_cellpose')
+reticulate::py_install(
+ pip = TRUE,
+ envname = 'giotto_cellpose',
+ packages = c(
+ "pandas",
+ "networkx",
+ "python-igraph",
+ "leidenalg",
+ "scikit-learn",
+ "cellpose",
+ "smfishhmrf",
+ 'tifffile',
+ 'scikit-image'
+ )
+)
+#.rs.restartR()
Now, activate the Giotto python environment.
-#.rs.restartR()
-# Activate the Giotto python environment of your choice
-GiottoClass::set_giotto_python_path('giotto_cellpose')
-# Check if cellpose was successfully installed
-GiottoUtils::package_check('cellpose',repository = 'pip')
+#.rs.restartR()
+# Activate the Giotto python environment of your choice
+GiottoClass::set_giotto_python_path('giotto_cellpose')
+# Check if cellpose was successfully installed
+GiottoUtils::package_check('cellpose',repository = 'pip')
11.1 Spatial Proteomics Technologies
This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix.
@@ -557,58 +557,58 @@
11.2 Raw data type coming out of
11.2.1 Use ome.tiff as an example output data to begin with
OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore.
-library(Giotto)
-instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
- save_plot = TRUE,
- show_plot = TRUE,
- python_path = 'giotto_cellpose')
-options(timeout = 9999999)
-download_dir =file.path(getwd(),'/data/02_session4/')
-destfile = file.path(download_dir,'Lunaphore.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
-list.files(paste0(download_dir,'/Lunaphore'))
+library(Giotto)
+instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ python_path = 'giotto_cellpose')
+options(timeout = 9999999)
+download_dir =file.path(getwd(),'/data/02_session4/')
+destfile = file.path(download_dir,'Lunaphore.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
+list.files(paste0(download_dir,'/Lunaphore'))
We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML.
-img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
-img_meta
+img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
+img_meta
However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it.
-##Get channel information
-Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
-node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
-channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
-channel_df
+##Get channel information
+Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
+node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
+channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
+channel_df
11.2.2 Use single channel images as an example output data to begin with
Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like.
-#Create multichannel raster and extract each single channels
-Luna_terra = terra::rast(Luna)
-names(Luna_terra) <- channel_df$Name
-gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
-plot(gimg_DAPI)
+#Create multichannel raster and extract each single channels
+Luna_terra = terra::rast(Luna)
+names(Luna_terra) <- channel_df$Name
+gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
+plot(gimg_DAPI)
Extract and save the raster image for future use.
-single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
-if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
-for (i in 1:nrow(channel_df)){
- single_channel <- terra::subset(Luna_terra, i)
- terra::writeRaster(single_channel,
- filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
- overwrite=T)
-}
+single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
+if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
+for (i in 1:nrow(channel_df)){
+ single_channel <- terra::subset(Luna_terra, i)
+ terra::writeRaster(single_channel,
+ filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
+ overwrite=T)
+}
Create a list of GiottoLargeImages using single channel rasters.
-file_names = list.files(single_channel_dir,full.names = TRUE)
-image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
-gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
- names = image_names,
- negative_y = F,
- flip_vertical = T)
-names(gimg_list) <- image_names
-plot(gimg_list[['Vimentin']])
+file_names = list.files(single_channel_dir,full.names = TRUE)
+image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
+gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
+ names = image_names,
+ negative_y = F,
+ flip_vertical = T)
+names(gimg_list) <- image_names
+plot(gimg_list[['Vimentin']])
@@ -618,34 +618,34 @@ 11.3 Cell Segmentation file to ge
11.3.1 Using segmentation output file from DeepCell(mesmer) as an example.
We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function.
-gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
-plot(gpoly_mesmer)
+gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
+plot(gpoly_mesmer)
We can also zoom in to check how does the segmentation look.
-zoom <- c(2000,2500,2000,2500)
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+zoom <- c(2000,2500,2000,2500)
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
11.3.2 Using Giotto wrapper of Cellpose to perform segmentation
-gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
-writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
-#Create a giotto image to evaluate segmentation
-gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
+gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
+writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
+#Create a giotto image to evaluate segmentation
+gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website
-doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
- mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- channel_1 = 0,
- channel_2 = 0,
- model_name = 'cyto3',
- batch_size=12)
-cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- shift_horizontal_step = F,
- shift_vertical_step = F,
- flip_vertical = T)
-plot(gimg_for_cellpose)
-plot(cpoly, add = T, border = 'red')
+doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
+ mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ channel_1 = 0,
+ channel_2 = 0,
+ model_name = 'cyto3',
+ batch_size=12)
+cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ shift_horizontal_step = F,
+ shift_vertical_step = F,
+ flip_vertical = T)
+plot(gimg_for_cellpose)
+plot(cpoly, add = T, border = 'red')
@@ -654,133 +654,133 @@ 11.4 Create a Giotto Object using
You will need to have
- list of giotto images
- giottoPolygon created from segmentation
-Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
- images = gimg_list,
- instructions = instrs)
-Lunaphore_giotto
+Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
+ images = gimg_list,
+ instructions = instrs)
+Lunaphore_giotto
11.4.1 Overlap to matrix
calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with
-Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
- spatial_info = 'cell',
- image_names = names(gimg_list))
-
-Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
- type ='intensity',
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-showGiottoExpression(Lunaphore_giotto)
+Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
+ spatial_info = 'cell',
+ image_names = names(gimg_list))
+
+Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
+ type ='intensity',
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+showGiottoExpression(Lunaphore_giotto)
11.4.2 Manipulate Expression information
For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile.
Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object
-expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
-filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
-Lunaphore_giotto <- setExpression(Lunaphore_giotto,
- feat_type = 'protein',
- x = createExprObj(filtered_expr_mtx),
- name = 'raw')
-showGiottoExpression(Lunaphore_giotto)
+expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
+filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
+Lunaphore_giotto <- setExpression(Lunaphore_giotto,
+ feat_type = 'protein',
+ x = createExprObj(filtered_expr_mtx),
+ name = 'raw')
+showGiottoExpression(Lunaphore_giotto)
11.4.3 Rescale polygons
rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis
-Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
- poly_info = 'cell',
- name = 'smallcell',
- fx = 0.7, fy = 0.7, calculate_centroids = T)
-smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
-plot(smallpoly,add = TRUE, border = "red",ext = zoom)
+Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
+ poly_info = 'cell',
+ name = 'smallcell',
+ fx = 0.7, fy = 0.7, calculate_centroids = T)
+smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+plot(smallpoly,add = TRUE, border = "red",ext = zoom)
11.4.4 Perform clustering and differential expression
The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering
-Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
-Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
-
-Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
- scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
- set_seed = TRUE)
-screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
+Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
+Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
+
+Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
+ scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
+ set_seed = TRUE)
+screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check.
-Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
- dimensions_to_use = 1:5,
- set_seed = TRUE)
-Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
-Lunaphore_giotto = doHclust(Lunaphore_giotto,
- k = 8,
- dim_reduction_to_use = 'cells',
- spat_unit = 'cell',
- feat_type = 'protein')
-
-spatInSituPlotPoints(gobject = Lunaphore_giotto,
- spat_unit = "cell",
- polygon_feat_type = "cell",
- show_polygon = T,
- feat_type = "protein",
- feats = NULL,
- polygon_fill = 'hclust',
- polygon_fill_as_factor = TRUE,
- polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
- polygon_color = 'black',
- background_color = "white")
+Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
+ dimensions_to_use = 1:5,
+ set_seed = TRUE)
+Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
+Lunaphore_giotto = doHclust(Lunaphore_giotto,
+ k = 8,
+ dim_reduction_to_use = 'cells',
+ spat_unit = 'cell',
+ feat_type = 'protein')
+
+spatInSituPlotPoints(gobject = Lunaphore_giotto,
+ spat_unit = "cell",
+ polygon_feat_type = "cell",
+ show_polygon = T,
+ feat_type = "protein",
+ feats = NULL,
+ polygon_fill = 'hclust',
+ polygon_fill_as_factor = TRUE,
+ polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
+ polygon_color = 'black',
+ background_color = "white")
Then we can check the heatmap of protein expression and determine the first round of cluster annotation
-cluster_column = 'hclust'
-plotMetaDataHeatmap(Lunaphore_giotto,
- spat_unit = 'cell',feat_type = 'protein',
- expression_values = "raw",
- metadata_cols = c(cluster_column),
- selected_feats = names(gimg_list),
- y_text_size = 8,
- show_values = 'zscores_rescaled')
+cluster_column = 'hclust'
+plotMetaDataHeatmap(Lunaphore_giotto,
+ spat_unit = 'cell',feat_type = 'protein',
+ expression_values = "raw",
+ metadata_cols = c(cluster_column),
+ selected_feats = names(gimg_list),
+ y_text_size = 8,
+ show_values = 'zscores_rescaled')
11.4.5 Give the cluster an annotation based on expression values
-
+
11.4.6 spatial network
This is to create a cellular neighborhood based on nearest neighbor of physical distance
-Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
-
-spatPlot2D(Lunaphore_giotto,
- show_network = T,
- network_color = 'blue',
- point_size = 1.5,
- cell_color = 'hclust')
+Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
+
+spatPlot2D(Lunaphore_giotto,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1.5,
+ cell_color = 'hclust')
11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions
This is using cellProximityEnrichment() to statistically identify cell type interactions
-cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
- cluster_column = 'cell_types',
- spatial_network_name = 'Delaunay_network',
- adjust_method = 'fdr',
- number_of_simulations = 2000)
-## barplot
-cellProximityBarplot(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- min_orig_ints = 5, min_sim_ints = 5)
+cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
+ cluster_column = 'cell_types',
+ spatial_network_name = 'Delaunay_network',
+ adjust_method = 'fdr',
+ number_of_simulations = 2000)
+## barplot
+cellProximityBarplot(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ min_orig_ints = 5, min_sim_ints = 5)
-## network
-cellProximityNetwork(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- remove_self_edges = T,
- only_show_enrichment_edges = F)
+## network
+cellProximityNetwork(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ remove_self_edges = T,
+ only_show_enrichment_edges = F)
diff --git a/visium-hd.html b/visium-hd.html
index 23aa210..f24cdd4 100644
--- a/visium-hd.html
+++ b/visium-hd.html
@@ -525,7 +525,7 @@ 9.1 Objective9.2 Background
9.2.1 Visium HD Technology
-
+
Figure 9.1: Overview of Visium HD. Source: 10X Genomics
@@ -536,7 +536,7 @@
9.2.1 Visium HD Technology
9.2.2 Colorectal Cancer Sample
-
+
Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics
@@ -550,7 +550,7 @@
9.2.2 Colorectal Cancer Sample9.3 Data Ingestion
9.3.1 Visium HD output data format
-
+
Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline.
@@ -560,19 +560,19 @@
9.3.1 Visium HD output data forma
9.4 Tiling and aggregation
The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells.
-
+
9.5 Scalability and projection functions
diff --git a/visium-part-i.html b/visium-part-i.html
index 5aa1658..c4746ee 100644
--- a/visium-part-i.html
+++ b/visium-part-i.html
@@ -520,7 +520,7 @@ 7 Visium Part I
7.1 The Visium technology
Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression.
-
+
Figure 7.1: Visum workflow. Source: 10X Genomics
@@ -536,73 +536,73 @@
7.2 Introduction to the spatial d
7.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/01_session5/")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
+dir.create("data/01_session5/")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”.
-
+
7.4 Create the Giotto object
createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/01_session5/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-## Provide the path to the visium folder
-data_path <- "data/01_session5/"
-
-## Create object directly from the visium folder
-visium_brain <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/01_session5/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+## Provide the path to the visium folder
+data_path <- "data/01_session5/"
+
+## Create object directly from the visium folder
+visium_brain <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
7.5 Subset on spots that were covered by tissue
Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area.
-spatPlot2D(
- gobject = visium_brain,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE)
-
+spatPlot2D(
+ gobject = visium_brain,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE)
+
Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area.
-
+
7.6 Quality control
@@ -610,43 +610,43 @@ 7.6 Quality controlvisium_brain_statistics <- addStatistics(gobject = visium_brain,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = visium_brain_statistics,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain_statistics <- addStatistics(gobject = visium_brain,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = visium_brain_statistics,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.3: Spatial distribution of features per spot.
filterDistributions() creates a histogram to show the distribution of features per spot across the sample.
-
-
+
+
Figure 7.4: Distribution of features per spot.
When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample.
-
-
+
+
Figure 7.5: Distribution of cells with different features per spot.
filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data:
-filterCombinations(gobject = visium_brain_statistics,
- expression_thresholds = c(1, 2, 3),
- feat_det_in_min_cells = c(50, 100, 200),
- min_det_feats_per_cell = c(500, 1000, 1500))
-
+filterCombinations(gobject = visium_brain_statistics,
+ expression_thresholds = c(1, 2, 3),
+ feat_det_in_min_cells = c(50, 100, 200),
+ min_det_feats_per_cell = c(500, 1000, 1500))
+
Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations.
@@ -656,34 +656,34 @@
7.6 Quality control
7.7 Filtering
Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample.
-visium_brain <- filterGiotto(
- gobject = visium_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE
-)
-Feature type: rna
-Number of cells removed: 4 out of 2702
-Number of feats removed: 7311 out of 22125
+
+
7.8 Normalization
Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one.
-visium_brain <- normalizeGiotto(
- gobject = visium_brain,
- scalefactor = 6000,
- verbose = TRUE
-)
+visium_brain <- normalizeGiotto(
+ gobject = visium_brain,
+ scalefactor = 6000,
+ verbose = TRUE
+)
Calculate the normalized number of features per spot and save the statistics in the metadata table.
-visium_brain <- addStatistics(gobject = visium_brain)
-
-## visualize
-spatPlot2D(gobject = visium_brain,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain <- addStatistics(gobject = visium_brain)
+
+## visualize
+spatPlot2D(gobject = visium_brain,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.7: Spatial distribution of the number of features per spot.
@@ -698,11 +698,11 @@
7.9.1 Highly Variable Features:
loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_loess",
- save_plot = TRUE,
- default_save_name = "HVFplot_loess")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_loess",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_loess")
+
Figure 7.8: Covariance of HVFs using the loess method.
@@ -711,11 +711,11 @@
7.9.1 Highly Variable Features:
pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")
+
Figure 7.9: Variance of HVFs using the pearson residuals method.
@@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
17.2 Downloading the dataset
We first need to import a dataset that we want to perform kriging on.
-data_directory <- "data"
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz")data_directory <- "data"
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz")After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”.
17.3 Importing visium data
We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network.
-library(Giotto)
-
-save_directory <- 'results'
-visium_save_directory <- file.path(save_directory, 'visium_mouse_brain')
-subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/')
-
-instrs <- createGiottoInstructions(show_plot = TRUE,
- save_plot = TRUE,
- save_dir = visium_save_directory)
-
-v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs)
-
-# Subset to in tissue only
-cm = pDataDT(v_brain)
-in_tissue_barcodes = cm[in_tissue == 1]$cell_ID
-v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes)
-
-# Filter
-v_brain = filterGiotto(gobject = v_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = c('raw'))
-
-# Normalize
-v_brain = normalizeGiotto(gobject = v_brain,
- scalefactor = 6000,
- verbose = TRUE)
-
-# Add stats
-v_brain = addStatistics(gobject = v_brain)
-
-# ID HVF
-v_brain = calculateHVF(gobject = v_brain, method = "cov_loess")
-fm = fDataDT(v_brain)
-hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID
-length(hv_feats)
-
-# Dimension Reductions
-v_brain = runPCA(gobject = v_brain,
- feats_to_use = hv_feats)
-
-v_brain = runUMAP(v_brain,
- dimensions_to_use = 1:10,
- n_neighbors = 15,
- set_seed = TRUE)
-
-# NN Network
-v_brain = createNearestNetwork(gobject = v_brain,
- dimensions_to_use = 1:10,
- k = 15)
-# Leiden Cluster
-v_brain = doLeidenCluster(gobject = v_brain,
- resolution = 0.4,
- n_iterations = 1000,
- set_seed = TRUE)
-
-# Spatial Network (kNN)
-v_brain <- createSpatialNetwork(gobject = v_brain,
- method = 'kNN',
- k = 5,
- maximum_distance_knn = 400,
- name = 'spatial_network')
-
-spatPlot2D(gobject = v_brain,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 1.5,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = T,
- save_param = list(save_name = '03_ses6_1_vis_spat'))library(Giotto)
+
+save_directory <- 'results'
+visium_save_directory <- file.path(save_directory, 'visium_mouse_brain')
+subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/')
+
+instrs <- createGiottoInstructions(show_plot = TRUE,
+ save_plot = TRUE,
+ save_dir = visium_save_directory)
+
+v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs)
+
+# Subset to in tissue only
+cm = pDataDT(v_brain)
+in_tissue_barcodes = cm[in_tissue == 1]$cell_ID
+v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes)
+
+# Filter
+v_brain = filterGiotto(gobject = v_brain,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1000,
+ expression_values = c('raw'))
+
+# Normalize
+v_brain = normalizeGiotto(gobject = v_brain,
+ scalefactor = 6000,
+ verbose = TRUE)
+
+# Add stats
+v_brain = addStatistics(gobject = v_brain)
+
+# ID HVF
+v_brain = calculateHVF(gobject = v_brain, method = "cov_loess")
+fm = fDataDT(v_brain)
+hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID
+length(hv_feats)
+
+# Dimension Reductions
+v_brain = runPCA(gobject = v_brain,
+ feats_to_use = hv_feats)
+
+v_brain = runUMAP(v_brain,
+ dimensions_to_use = 1:10,
+ n_neighbors = 15,
+ set_seed = TRUE)
+
+# NN Network
+v_brain = createNearestNetwork(gobject = v_brain,
+ dimensions_to_use = 1:10,
+ k = 15)
+# Leiden Cluster
+v_brain = doLeidenCluster(gobject = v_brain,
+ resolution = 0.4,
+ n_iterations = 1000,
+ set_seed = TRUE)
+
+# Spatial Network (kNN)
+v_brain <- createSpatialNetwork(gobject = v_brain,
+ method = 'kNN',
+ k = 5,
+ maximum_distance_knn = 400,
+ name = 'spatial_network')
+
+spatPlot2D(gobject = v_brain,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 1.5,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = T,
+ save_param = list(save_name = '03_ses6_1_vis_spat'))
+
17.3 Importing visium data
Figure 17.2: Mouse brain spatial plot showing leiden clustering @@ -638,17 +638,17 @@
17.3 Importing visium data
17.3.1 Identifying spatially organized features
We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time.
-# Spatially Variable Features
-ranktest = binSpect(v_brain,
- bin_method = 'rank',
- calc_hub = T,
- hub_min_int = 5,
- spatial_network_name = 'spatial_network',
- do_parallel = T,
- cores = 8) #not able to provide a seed number, so do not set one
-
-# Getting the top 15 spatially organized genes
-ext_spatial_features = ranktest[1:15,]$feats
+# Spatially Variable Features
+ranktest = binSpect(v_brain,
+ bin_method = 'rank',
+ calc_hub = T,
+ hub_min_int = 5,
+ spatial_network_name = 'spatial_network',
+ do_parallel = T,
+ cores = 8) #not able to provide a seed number, so do not set one
+
+# Getting the top 15 spatially organized genes
+ext_spatial_features = ranktest[1:15,]$feats
# Spatially Variable Features
-ranktest = binSpect(v_brain,
- bin_method = 'rank',
- calc_hub = T,
- hub_min_int = 5,
- spatial_network_name = 'spatial_network',
- do_parallel = T,
- cores = 8) #not able to provide a seed number, so do not set one
-
-# Getting the top 15 spatially organized genes
-ext_spatial_features = ranktest[1:15,]$feats# Spatially Variable Features
+ranktest = binSpect(v_brain,
+ bin_method = 'rank',
+ calc_hub = T,
+ hub_min_int = 5,
+ spatial_network_name = 'spatial_network',
+ do_parallel = T,
+ cores = 8) #not able to provide a seed number, so do not set one
+
+# Getting the top 15 spatially organized genes
+ext_spatial_features = ranktest[1:15,]$feats
@@ -656,27 +656,27 @@ 17.4 Adding cell polygons to Giot
17.4 Adding cell polygons to Giot
17.4.1 Read in the poly information
First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object.
-# Read in the data
-stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson")
-
-stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path,
- name = "stardist_cell",
- calc_centroids = TRUE)
-
-stardist_cell_gpoly <- flip(stardist_cell_gpoly)
+# Read in the data
+stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson")
+
+stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path,
+ name = "stardist_cell",
+ calc_centroids = TRUE)
+
+stardist_cell_gpoly <- flip(stardist_cell_gpoly)
17.4.2 Vizualizing polygons
Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain.
-
-
+
+
Figure 17.3: Mouse brain cell polygons from the visium dataset
-
+
Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped
@@ -686,8 +686,8 @@
17.4.2 Vizualizing polygons
17.4.3 Showing Giotto object prior to polygon addition
Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon.
-
-
+
+
Figure 17.5: Giotto object before adding subcellular polygons.
@@ -697,11 +697,11 @@
17.4.3 Showing Giotto object prio
17.4.4 Adding polygons to giotto object
After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject.
-v_brain <- addGiottoPolygons(v_brain,
- gpolygons = list('stardist_cell' = stardist_cell_gpoly))
-
-print(v_brain)
-
+v_brain <- addGiottoPolygons(v_brain,
+ gpolygons = list('stardist_cell' = stardist_cell_gpoly))
+
+print(v_brain)
+
Figure 17.6: Giotto object after to adding subcellular polygons.
@@ -711,9 +711,9 @@
17.4.4 Adding polygons to giotto
17.4.5 Check polygon information
We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots.
-
-
+
+
Figure 17.7: Polygon information for stardist_cell.
@@ -727,23 +727,23 @@
17.5 Performing kriging17.5.1 Interpolating features
Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be
analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required.
-future::plan(future::multisession()) # comment out for single threading
-subcell_v_brain <- interpolateFeature(v_brain,
- spat_unit = 'cell',
- feat_type = 'rna',
- ext = ext(v_brain),
- feats = ext_spatial_features,
- overwrite = T)
-
-print(subcell_v_brain)
-
+future::plan(future::multisession()) # comment out for single threading
+subcell_v_brain <- interpolateFeature(v_brain,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ ext = ext(v_brain),
+ feats = ext_spatial_features,
+ overwrite = T)
+
+print(subcell_v_brain)
+
Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right).
For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1.
-
+
Figure 17.9: Raster of gene expression interpolation for Pantr1
@@ -754,18 +754,18 @@
17.5.1 Interpolating features17.5.2 Expression overlap
The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier.
This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run.
-subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain,
- name_overlap = "rna",
- spatial_info = "stardist_cell",
- image_names = ext_spatial_features)
-
-subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain,
- poly_info = "stardist_cell",
- feat_info = "rna",
- aggr_function = "sum",
- type='intensity')
+subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain,
+ name_overlap = "rna",
+ spatial_info = "stardist_cell",
+ image_names = ext_spatial_features)
+
+subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain,
+ poly_info = "stardist_cell",
+ feat_info = "rna",
+ aggr_function = "sum",
+ type='intensity')
After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified.
-
+
Figure 17.10: Gene expression for cells based on interpolation.
@@ -776,70 +776,70 @@
17.5.2 Expression overlap
17.6 Reading in larger dataset
For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here.
-
+
17.7 Analyzing interpolated features
17.7.1 Filter and normalization
Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis.
-subcell_v_brain <- filterGiotto(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- expression_values = "raw",
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-
-subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- scalefactor = 6000,
- verbose = T)
+subcell_v_brain <- filterGiotto(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ expression_values = "raw",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+
+subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ scalefactor = 6000,
+ verbose = T)
17.7.2 Visualizing gene expression from interpolated expression
Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data.
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Sparc',
- point_size = 2,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_sparc_vis'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'stardist_cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Sparc',
- point_size = 0.6,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_sparc'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'cell',
- gradient_style = 'sequential',
- feats = 'Pantr1',
- cell_color_gradient = 'Geyser',
- point_size = 2,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_pantr1_vis'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'stardist_cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Pantr1',
- point_size = 0.6,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_pantr1'))
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Sparc',
+ point_size = 2,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_sparc_vis'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Sparc',
+ point_size = 0.6,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_sparc'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'cell',
+ gradient_style = 'sequential',
+ feats = 'Pantr1',
+ cell_color_gradient = 'Geyser',
+ point_size = 2,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_pantr1_vis'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Pantr1',
+ point_size = 0.6,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_pantr1'))
Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer.
-
+
Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom).
@@ -848,37 +848,37 @@
17.7.2 Visualizing gene expressio
17.7.3 Run PCA
-
+
17.7.4 Clustering
-# UMAP
-subcell_v_brain <- runUMAP(subcell_v_brain,
- spat_unit = "stardist_cell",
- dimensions_to_use = 1:15,
- n_neighbors = 1000,
- min_dist = 0.001,
- spread = 1)
-
-# NN Network
-subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- dimensions_to_use = 1:10,
- feats_to_use = hv_feats,
- expression_values = 'normalized',
- k = 70)
-
-subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- resolution = 0.15,
- n_iterations = 100,
- partition_type = 'RBConfigurationVertexPartition')
-
-plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus')
-
+# UMAP
+subcell_v_brain <- runUMAP(subcell_v_brain,
+ spat_unit = "stardist_cell",
+ dimensions_to_use = 1:15,
+ n_neighbors = 1000,
+ min_dist = 0.001,
+ spread = 1)
+
+# NN Network
+subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ dimensions_to_use = 1:10,
+ feats_to_use = hv_feats,
+ expression_values = 'normalized',
+ k = 70)
+
+subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ resolution = 0.15,
+ n_iterations = 100,
+ partition_type = 'RBConfigurationVertexPartition')
+
+plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus')
+
Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering.
@@ -888,27 +888,27 @@
17.7.4 Clustering
17.7.5 Visualizing clustering
Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster.
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.5,
- point_shape = 'no_border',
- background_color = 'black',
- save_plot = F,
- show_legend = TRUE)
-
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat'))
-
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.5,
+ point_shape = 'no_border',
+ background_color = 'black',
+ save_plot = F,
+ show_legend = TRUE)
+
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat'))
+
Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
@@ -918,37 +918,37 @@
17.7.5 Visualizing clustering
17.7.6 Cropping objects
We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below.
-subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
- spat_unit = ":all:",
- x_min = 4000,
- x_max = 7000,
- y_min = -6500,
- y_max = -3500,
- z_max = NULL,
- z_min = NULL)
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 2,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_vis_spat_crop'))
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat_crop'))
-
+subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
+ spat_unit = ":all:",
+ x_min = 4000,
+ x_max = 7000,
+ y_min = -6500,
+ y_max = -3500,
+ z_max = NULL,
+ z_min = NULL)
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 2,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_vis_spat_crop'))
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat_crop'))
+
Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
diff --git a/interoperability-with-isolated-tools.html b/interoperability-with-isolated-tools.html
index ceb2d0a..8cca7e6 100644
--- a/interoperability-with-isolated-tools.html
+++ b/interoperability-with-isolated-tools.html
@@ -526,1521 +526,1521 @@
16.1.1 Dataset downloaddata_path <- "data"
-dir.create(data_path)
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
- destfile = file.path(data_path,"counts.h5ad"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
- destfile = file.path(data_path,"cell_labels.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
+data_path <- "data"
+dir.create(data_path)
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
+ destfile = file.path(data_path,"counts.h5ad"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
+ destfile = file.path(data_path,"cell_labels.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
16.1.2 Create the Giotto object
-library(Giotto)
-library(reticulate)
-
-## Set instructions
-results_folder <- "results"
-dir.create(results_folder)
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-
-## load data
-
-### meta data
-meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-colnames(meta_df)[[1]] <- "cell_ID"
-slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
-slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
-
-### counts matrix
-scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
-ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
-reticulate::source_python(ad2g_path)
-adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
-X <- extract_expression(adata)
-cID <- extract_cell_IDs(adata)
-fID <- extract_feat_IDs(adata)
-rownames(X) <- fID
-colnames(X) <- cID
-slice1_filter <- cID %in% slice1_meta_df$cell_ID
-slice2_filter <- cID %in% slice2_meta_df$cell_ID
-slice1_exp <- X[,slice1_filter]
-slice2_exp <- X[,slice2_filter]
-
-### cell segmentation to cell location
-segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_df <- rbind(segments_1_df, segments_2_df)
-loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
-loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
-loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
-centr.x <- apply(loc.use[,loc.x],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-centr.y <- apply(loc.use[,loc.y],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
-slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
-slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
-
-## giotto obj
-giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
- spatial_locs = slice1_spatial_locs_df,
- cell_metadata = slice1_meta_df)
-giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
- spatial_locs = slice2_spatial_locs_df,
- cell_metadata = slice2_meta_df)
-giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
- gobject_names = c('mouse2_slice229',
- 'mouse2_slice300'), # name for each samples
- join_method = 'z_stack')
-# We skip the processing process here to save time and use the given cell type
-# annotation directly
-
-ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
- cell_type = "subclass",
- output_path = data_path,
- spat_unit = "cell",
- feat_type = "rna",
- verbose = TRUE)
-
-# Cell_ID Sample x y Cell_Type
-# <char> <char> <dbl> <dbl> <char>
-# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
-# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
-# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
-# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
-# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
-# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
+library(Giotto)
+library(reticulate)
+
+## Set instructions
+results_folder <- "results"
+dir.create(results_folder)
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+
+## load data
+
+### meta data
+meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+colnames(meta_df)[[1]] <- "cell_ID"
+slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
+slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
+
+### counts matrix
+scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
+ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
+reticulate::source_python(ad2g_path)
+adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
+X <- extract_expression(adata)
+cID <- extract_cell_IDs(adata)
+fID <- extract_feat_IDs(adata)
+rownames(X) <- fID
+colnames(X) <- cID
+slice1_filter <- cID %in% slice1_meta_df$cell_ID
+slice2_filter <- cID %in% slice2_meta_df$cell_ID
+slice1_exp <- X[,slice1_filter]
+slice2_exp <- X[,slice2_filter]
+
+### cell segmentation to cell location
+segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_df <- rbind(segments_1_df, segments_2_df)
+loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
+loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
+loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
+centr.x <- apply(loc.use[,loc.x],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+centr.y <- apply(loc.use[,loc.y],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
+slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
+slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
+
+## giotto obj
+giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
+ spatial_locs = slice1_spatial_locs_df,
+ cell_metadata = slice1_meta_df)
+giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
+ spatial_locs = slice2_spatial_locs_df,
+ cell_metadata = slice2_meta_df)
+giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
+ gobject_names = c('mouse2_slice229',
+ 'mouse2_slice300'), # name for each samples
+ join_method = 'z_stack')
+# We skip the processing process here to save time and use the given cell type
+# annotation directly
+
+ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
+ cell_type = "subclass",
+ output_path = data_path,
+ spat_unit = "cell",
+ feat_type = "rna",
+ verbose = TRUE)
+
+# Cell_ID Sample x y Cell_Type
+# <char> <char> <dbl> <dbl> <char>
+# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
+# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
+# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
+# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
+# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
+# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
Spatial cell type distributions
-spatPlot2D(giotto_obj,
- group_by = "slice_id",
- cell_color = "subclass",
- point_size = 1,
- point_border_stroke = NA,
- legend_text = 6)
+spatPlot2D(giotto_obj,
+ group_by = "slice_id",
+ cell_color = "subclass",
+ point_size = 1,
+ point_border_stroke = NA,
+ legend_text = 6)
16.1.2.1 Installation ONTraC
You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node.
It will run for less than 10 minutes on this example dataset.
For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time.
-source ~/.bash_profile
-conda create -y -n ONTraC python=3.11
-conda activate ONTraC
-pip install git+https://github.com/gyuanlab/ONTraC.git@V2
-# Retrieving notices: ...working... done
-# Channels:
-# - conda-forge
-# - bioconda
-# - r
-# - pytorch
-# - defaults
-# Platform: osx-arm64
-# Collecting package metadata (repodata.json): ...working... done
-# Solving environment: ...working... done
-#
-# ## Package Plan ##
-#
-# environment location: /Users/wwang/miniconda3/envs/ONTraC
-#
-# added / updated specs:
-# - python=3.11
-#
-#
-# The following packages will be downloaded:
-#
-# package | build
-# ---------------------------|-----------------
-# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
-# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
-# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
-# ------------------------------------------------------------
-# Total: 4.3 MB
-#
-# The following NEW packages will be INSTALLED:
-#
-# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
-# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
-# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
-# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
-# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
-# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
-# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
-# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
-# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
-# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
-# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
-# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
-# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
-# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
-# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
-# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
-#
-#
-#
-# Downloading and Extracting Packages: ...working... done
-# Preparing transaction: ...working... done
-# Verifying transaction: ...working... done
-# Executing transaction: ...working... done
-# #
-# # To activate this environment, use
-# #
-# # $ conda activate ONTraC
-# #
-# # To deactivate an active environment, use
-# #
-# # $ conda deactivate
-#
-# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
-# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
-# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
-# Running command git checkout -b V2 --track origin/V2
-# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
-# Installing build dependencies: started
-# Switched to a new branch 'V2'
-# branch 'V2' set up to track 'origin/V2'.
-# Installing build dependencies: finished with status 'done'
-# Getting requirements to build wheel: started
-# Getting requirements to build wheel: finished with status 'done'
-# Preparing metadata (pyproject.toml): started
-# Preparing metadata (pyproject.toml): finished with status 'done'
-# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
-# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
-# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
-# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
-# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
-# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
-# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
-# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
-# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
-# Collecting session-info (from ONTraC==2.0a9.dev7)
-# Using cached session_info-1.0.0-py3-none-any.whl
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
-# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
-# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
-# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
-# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
-# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
-# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
-# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
-# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
-# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
-# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
-# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
-# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
-# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
-# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
-# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
-# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
-# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
-# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
-# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
-# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
-# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
-# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
-# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
-# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
-# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
-# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
-# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
-# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
-# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
-# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
-# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
-# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
-# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
-# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
-# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
-# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
-# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
-# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
-# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
-# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
-# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
-# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
-# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
-# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
-# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
-# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
-# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
-# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
-# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
-# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
-# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
-# Using cached idna-3.7-py3-none-any.whl (66 kB)
-# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
-# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
-# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
-# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
-# Building wheels for collected packages: ONTraC
-# Building wheel for ONTraC (pyproject.toml): started
-# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
-# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
-# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
-# Successfully built ONTraC
-# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
-# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
+source ~/.bash_profile
+conda create -y -n ONTraC python=3.11
+conda activate ONTraC
+pip install git+https://github.com/gyuanlab/ONTraC.git@V2
+# Retrieving notices: ...working... done
+# Channels:
+# - conda-forge
+# - bioconda
+# - r
+# - pytorch
+# - defaults
+# Platform: osx-arm64
+# Collecting package metadata (repodata.json): ...working... done
+# Solving environment: ...working... done
+#
+# ## Package Plan ##
+#
+# environment location: /Users/wwang/miniconda3/envs/ONTraC
+#
+# added / updated specs:
+# - python=3.11
+#
+#
+# The following packages will be downloaded:
+#
+# package | build
+# ---------------------------|-----------------
+# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
+# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
+# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
+# ------------------------------------------------------------
+# Total: 4.3 MB
+#
+# The following NEW packages will be INSTALLED:
+#
+# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
+# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
+# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
+# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
+# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
+# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
+# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
+# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
+# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
+# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
+# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
+# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
+# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
+# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
+# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
+# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
+#
+#
+#
+# Downloading and Extracting Packages: ...working... done
+# Preparing transaction: ...working... done
+# Verifying transaction: ...working... done
+# Executing transaction: ...working... done
+# #
+# # To activate this environment, use
+# #
+# # $ conda activate ONTraC
+# #
+# # To deactivate an active environment, use
+# #
+# # $ conda deactivate
+#
+# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
+# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
+# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
+# Running command git checkout -b V2 --track origin/V2
+# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
+# Installing build dependencies: started
+# Switched to a new branch 'V2'
+# branch 'V2' set up to track 'origin/V2'.
+# Installing build dependencies: finished with status 'done'
+# Getting requirements to build wheel: started
+# Getting requirements to build wheel: finished with status 'done'
+# Preparing metadata (pyproject.toml): started
+# Preparing metadata (pyproject.toml): finished with status 'done'
+# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
+# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
+# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
+# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
+# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
+# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
+# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
+# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
+# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
+# Collecting session-info (from ONTraC==2.0a9.dev7)
+# Using cached session_info-1.0.0-py3-none-any.whl
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
+# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
+# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
+# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
+# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
+# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
+# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
+# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
+# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
+# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
+# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
+# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
+# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
+# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
+# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
+# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
+# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
+# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
+# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
+# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
+# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
+# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
+# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
+# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
+# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
+# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
+# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
+# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
+# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
+# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
+# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
+# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
+# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
+# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
+# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
+# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
+# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
+# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
+# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
+# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
+# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
+# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
+# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
+# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
+# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
+# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
+# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
+# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
+# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
+# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
+# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
+# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
+# Using cached idna-3.7-py3-none-any.whl (66 kB)
+# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
+# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
+# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
+# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
+# Building wheels for collected packages: ONTraC
+# Building wheel for ONTraC (pyproject.toml): started
+# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
+# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
+# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
+# Successfully built ONTraC
+# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
+# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
16.1.2.2 Running ONTraC
-
-source ~/.bash_profile
-conda activate ONTraC_v2_py311
-ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
-# ##################################################################################
-#
-# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
-# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
-# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
-# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
-# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
-#
-# version: 2.0a11
-#
-# ##################################################################################
-# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
-# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
-# 11:33:23 --- INFO: -------- I/O options -------
-# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
-# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
-# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
-# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
-# 11:33:23 --- INFO: -------- preprocessing options -------
-# 11:33:23 --- INFO: resolution: 10.0
-# 11:33:23 --- INFO: -------- niche net constr options -------
-# 11:33:23 --- INFO: n_cpu: 4
-# 11:33:23 --- INFO: n_neighbors: 50
-# 11:33:23 --- INFO: n_local: 20
-# 11:33:23 --- INFO: embedding_adjust: False
-# 11:33:23 --- INFO: sigma: 1
-# 11:33:23 --- INFO: -------- train options -------
-# 11:33:23 --- INFO: device: cpu
-# 11:33:23 --- INFO: epochs: 1000
-# 11:33:23 --- INFO: batch_size: 0
-# 11:33:23 --- INFO: patience: 100
-# 11:33:23 --- INFO: min_delta: 0.001
-# 11:33:23 --- INFO: min_epochs: 50
-# 11:33:23 --- INFO: seed: 42
-# 11:33:23 --- INFO: lr: 0.03
-# 11:33:23 --- INFO: hidden_feats: 4
-# 11:33:23 --- INFO: k: 6
-# 11:33:23 --- INFO: modularity_loss_weight: 1.0
-# 11:33:23 --- INFO: purity_loss_weight: 100.0
-# 11:33:23 --- INFO: regularization_loss_weight: 0.1
-# 11:33:23 --- INFO: beta: 0.3
-# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
-# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
-# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
-# 11:33:23 --- INFO: ------------- Niche network construct ---------------
-# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
-# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
-# 11:33:28 --- INFO: Generating samples.yaml file.
-# 11:33:28 --- INFO: ------------ Niche network construct end ------------
-# 11:33:28 --- INFO: ------------------------ GNN ------------------------
-# 11:33:28 --- INFO: Loading dataset.
-# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
-# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
-# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
-# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
-# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
-# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
-# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
-# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
-# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
-# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
-# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
-# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
-# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
-# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
-# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
-# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
-# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
-# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
-# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
-# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
-# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
-# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
-# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
-# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
-# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
-# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
-# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
-# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
-# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
-# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
-# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
-# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
-# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
-# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
-# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
-# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
-# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
-# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
-# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
-# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
-# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
-# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
-# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
-# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
-# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
-# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
-# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
-# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
-# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
-# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
-# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
-# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
-# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
-# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
-# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
-# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
-# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
-# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
-# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
-# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
-# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
-# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
-# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
-# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
-# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
-# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
-# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
-# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
-# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
-# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
-# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
-# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
-# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
-# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
-# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
-# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
-# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
-# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
-# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
-# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
-# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
-# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
-# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
-# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
-# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
-# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
-# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
-# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
-# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
-# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
-# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
-# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
-# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
-# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
-# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
-# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
-# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
-# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
-# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
-# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
-# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
-# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
-# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
-# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
-# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
-# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
-# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
-# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
-# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
-# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
-# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
-# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
-# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
-# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
-# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
-# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
-# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
-# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
-# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
-# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
-# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
-# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
-# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
-# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
-# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
-# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
-# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
-# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
-# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
-# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
-# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
-# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
-# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
-# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
-# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
-# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
-# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
-# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
-# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
-# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
-# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
-# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
-# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
-# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
-# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
-# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
-# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
-# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
-# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
-# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
-# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
-# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
-# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
-# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
-# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
-# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
-# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
-# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
-# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
-# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
-# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
-# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
-# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
-# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
-# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
-# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
-# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
-# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
-# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
-# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
-# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
-# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
-# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
-# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
-# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
-# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
-# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
-# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
-# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
-# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
-# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
-# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
-# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
-# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
-# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
-# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
-# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
-# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
-# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
-# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
-# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
-# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
-# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
-# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
-# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
-# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
-# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
-# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
-# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
-# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
-# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
-# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
-# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
-# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
-# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
-# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
-# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
-# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
-# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
-# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
-# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
-# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
-# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
-# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
-# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
-# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
-# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
-# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
-# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
-# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
-# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
-# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
-# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
-# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
-# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
-# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
-# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
-# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
-# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
-# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
-# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
-# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
-# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
-# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
-# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
-# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
-# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
-# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
-# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
-# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
-# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
-# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
-# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
-# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
-# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
-# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
-# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
-# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
-# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
-# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
-# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
-# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
-# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
-# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
-# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
-# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
-# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
-# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
-# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
-# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
-# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
-# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
-# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
-# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
-# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
-# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
-# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
-# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
-# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
-# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
-# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
-# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
-# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
-# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
-# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
-# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
-# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
-# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
-# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
-# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
-# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
-# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
-# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
-# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
-# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
-# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
-# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
-# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
-# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
-# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
-# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
-# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
-# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
-# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
-# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
-# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
-# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
-# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
-# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
-# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
-# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
-# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
-# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
-# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
-# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
-# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
-# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
-# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
-# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
-# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
-# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
-# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
-# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
-# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
-# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
-# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
-# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
-# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
-# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
-# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
-# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
-# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
-# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
-# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
-# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
-# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
-# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
-# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
-# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
-# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
-# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
-# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
-# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
-# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
-# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
-# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
-# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
-# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
-# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
-# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
-# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
-# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
-# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
-# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
-# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
-# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
-# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
-# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
-# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
-# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
-# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
-# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
-# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
-# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
-# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
-# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
-# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
-# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
-# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
-# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
-# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
-# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
-# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
-# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
-# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
-# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
-# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
-# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
-# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
-# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
-# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
-# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
-# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
-# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
-# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
-# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
-# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
-# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
-# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
-# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
-# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
-# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
-# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
-# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
-# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
-# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
-# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
-# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
-# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
-# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
-# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
-# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
-# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
-# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
-# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
-# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
-# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
-# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
-# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
-# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
-# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
-# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
-# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
-# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
-# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
-# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
-# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
-# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
-# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
-# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
-# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
-# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
-# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
-# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
-# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
-# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
-# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
-# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
-# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
-# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
-# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
-# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
-# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
-# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
-# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
-# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
-# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
-# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
-# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
-# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
-# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
-# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
-# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
-# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
-# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
-# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
-# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
-# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
-# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
-# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
-# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
-# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
-# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
-# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
-# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
-# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
-# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
-# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
-# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
-# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
-# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
-# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
-# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
-# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
-# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
-# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
-# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
-# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
-# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
-# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
-# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
-# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
-# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
-# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
-# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
-# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
-# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
-# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
-# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
-# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
-# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
-# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
-# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
-# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
-# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
-# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
-# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
-# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
-# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
-# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
-# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
-# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
-# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
-# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
-# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
-# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
-# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
-# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
-# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
-# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
-# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
-# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
-# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
-# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
-# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
-# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
-# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
-# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
-# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
-# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
-# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
-# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
-# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
-# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
-# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
-# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
-# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
-# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
-# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
-# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
-# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
-# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
-# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
-# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
-# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
-# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
-# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
-# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
-# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
-# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
-# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
-# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
-# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
-# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
-# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
-# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
-# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
-# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
-# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
-# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
-# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
-# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
-# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
-# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
-# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
-# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
-# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
-# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
-# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
-# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
-# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
-# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
-# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
-# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
-# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
-# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
-# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
-# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
-# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
-# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
-# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
-# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
-# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
-# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
-# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
-# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
-# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
-# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
-# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
-# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
-# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
-# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
-# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
-# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
-# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
-# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
-# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
-# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
-# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
-# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
-# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
-# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
-# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
-# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
-# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
-# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
-# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
-# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
-# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
-# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
-# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
-# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
-# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
-# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
-# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
-# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
-# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
-# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
-# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
-# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
-# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
-# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
-# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
-# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
-# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
-# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
-# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
-# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
-# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
-# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
-# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
-# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
-# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
-# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
-# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
-# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
-# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
-# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
-# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
-# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
-# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
-# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
-# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
-# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
-# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
-# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
-# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
-# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
-# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
-# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
-# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
-# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
-# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
-# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
-# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
-# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
-# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
-# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
-# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
-# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
-# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
-# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
-# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
-# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
-# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
-# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
-# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
-# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
-# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
-# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
-# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
-# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
-# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
-# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
-# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
-# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
-# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
-# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
-# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
-# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
-# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
-# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
-# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
-# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
-# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
-# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
-# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
-# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
-# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
-# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
-# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
-# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
-# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
-# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
-# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
-# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
-# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
-# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
-# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
-# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
-# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
-# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
-# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
-# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
-# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
-# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
-# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
-# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
-# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
-# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
-# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
-# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
-# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
-# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
-# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
-# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
-# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
-# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
-# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
-# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
-# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
-# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
-# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
-# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
-# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
-# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
-# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
-# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
-# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
-# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
-# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
-# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
-# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
-# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
-# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
-# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
-# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
-# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
-# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
-# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
-# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
-# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
-# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
-# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
-# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
-# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
-# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
-# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
-# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
-# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
-# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
-# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
-# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
-# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
-# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
-# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
-# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
-# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
-# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
-# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
-# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
-# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
-# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
-# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
-# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
-# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
-# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
-# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
-# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
-# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
-# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
-# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
-# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
-# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
-# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
-# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
-# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
-# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
-# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
-# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
-# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
-# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
-# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
-# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
-# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
-# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
-# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
-# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
-# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
-# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
-# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
-# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
-# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
-# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
-# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
-# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
-# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
-# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
-# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
-# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
-# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
-# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
-# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
-# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
-# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
-# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
-# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
-# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
-# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
-# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
-# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
-# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
-# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
-# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
-# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
-# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
-# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
-# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
-# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
-# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
-# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
-# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
-# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
-# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
-# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
-# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
-# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
-# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
-# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
-# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
-# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
-# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
-# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
-# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
-# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
-# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
-# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
-# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
-# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
-# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
-# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
-# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
-# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
-# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
-# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
-# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
-# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
-# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
-# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
-# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
-# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
-# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
-# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
-# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
-# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
-# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
-# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
-# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
-# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
-# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
-# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
-# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
-# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
-# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
-# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
-# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
-# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
-# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
-# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
-# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
-# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
-# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
-# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
-# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
-# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
-# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
-# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
-# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
-# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
-# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
-# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
-# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
-# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
-# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
-# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
-# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
-# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
-# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
-# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
-# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
-# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
-# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
-# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
-# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
-# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
-# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
-# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
-# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
-# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
-# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
-# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
-# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
-# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
-# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
-# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
-# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
-# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
-# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
-# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
-# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
-# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
-# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
-# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
-# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
-# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
-# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
-# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
-# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
-# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
-# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
-# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
-# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
-# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
-# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
-# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
-# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
-# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
-# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
-# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
-# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
-# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
-# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
-# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
-# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
-# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
-# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
-# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
-# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
-# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
-# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
-# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
-# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
-# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
-# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
-# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
-# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
-# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
-# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
-# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
-# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
-# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
-# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
-# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
-# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
-# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
-# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
-# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
-# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
-# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
-# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
-# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
-# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
-# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
-# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
-# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
-# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
-# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
-# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
-# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
-# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
-# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
-# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
-# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
-# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
-# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
-# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
-# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
-# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
-# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
-# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
-# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
-# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
-# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
-# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
-# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
-# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
-# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
-# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
-# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
-# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
-# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
-# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
-# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
-# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
-# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
-# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
-# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
-# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
-# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
-# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
-# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
-# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
-# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
-# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
-# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
-# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
-# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
-# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
-# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
-# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
-# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
-# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
-# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
-# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
-# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
-# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
-# 11:38:37 --- INFO: Training process end.
-# 11:38:37 --- INFO: Evaluating process start.
-# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
-# 11:38:37 --- INFO: Evaluating process end.
-# 11:38:37 --- INFO: Predicting process start.
-# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
-# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
-# 11:38:38 --- INFO: Predicting process end.
-# 11:38:38 --- INFO: --------------------- GNN end ----------------------
-# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
-# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
-# 11:38:38 --- INFO: Loading dataset.
-# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:38:39 --- INFO: Calculating NTScore for each niche.
-# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
-# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
-# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
-# 11:38:39 --- INFO: Output NTScore tables.
-# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
+
+source ~/.bash_profile
+conda activate ONTraC_v2_py311
+ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
+# ##################################################################################
+#
+# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
+# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
+# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
+# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
+# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
+#
+# version: 2.0a11
+#
+# ##################################################################################
+# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
+# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
+# 11:33:23 --- INFO: -------- I/O options -------
+# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
+# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
+# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
+# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
+# 11:33:23 --- INFO: -------- preprocessing options -------
+# 11:33:23 --- INFO: resolution: 10.0
+# 11:33:23 --- INFO: -------- niche net constr options -------
+# 11:33:23 --- INFO: n_cpu: 4
+# 11:33:23 --- INFO: n_neighbors: 50
+# 11:33:23 --- INFO: n_local: 20
+# 11:33:23 --- INFO: embedding_adjust: False
+# 11:33:23 --- INFO: sigma: 1
+# 11:33:23 --- INFO: -------- train options -------
+# 11:33:23 --- INFO: device: cpu
+# 11:33:23 --- INFO: epochs: 1000
+# 11:33:23 --- INFO: batch_size: 0
+# 11:33:23 --- INFO: patience: 100
+# 11:33:23 --- INFO: min_delta: 0.001
+# 11:33:23 --- INFO: min_epochs: 50
+# 11:33:23 --- INFO: seed: 42
+# 11:33:23 --- INFO: lr: 0.03
+# 11:33:23 --- INFO: hidden_feats: 4
+# 11:33:23 --- INFO: k: 6
+# 11:33:23 --- INFO: modularity_loss_weight: 1.0
+# 11:33:23 --- INFO: purity_loss_weight: 100.0
+# 11:33:23 --- INFO: regularization_loss_weight: 0.1
+# 11:33:23 --- INFO: beta: 0.3
+# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
+# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
+# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
+# 11:33:23 --- INFO: ------------- Niche network construct ---------------
+# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
+# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
+# 11:33:28 --- INFO: Generating samples.yaml file.
+# 11:33:28 --- INFO: ------------ Niche network construct end ------------
+# 11:33:28 --- INFO: ------------------------ GNN ------------------------
+# 11:33:28 --- INFO: Loading dataset.
+# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
+# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
+# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
+# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
+# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
+# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
+# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
+# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
+# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
+# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
+# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
+# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
+# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
+# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
+# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
+# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
+# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
+# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
+# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
+# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
+# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
+# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
+# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
+# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
+# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
+# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
+# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
+# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
+# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
+# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
+# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
+# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
+# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
+# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
+# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
+# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
+# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
+# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
+# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
+# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
+# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
+# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
+# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
+# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
+# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
+# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
+# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
+# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
+# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
+# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
+# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
+# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
+# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
+# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
+# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
+# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
+# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
+# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
+# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
+# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
+# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
+# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
+# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
+# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
+# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
+# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
+# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
+# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
+# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
+# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
+# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
+# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
+# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
+# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
+# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
+# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
+# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
+# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
+# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
+# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
+# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
+# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
+# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
+# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
+# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
+# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
+# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
+# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
+# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
+# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
+# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
+# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
+# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
+# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
+# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
+# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
+# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
+# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
+# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
+# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
+# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
+# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
+# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
+# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
+# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
+# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
+# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
+# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
+# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
+# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
+# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
+# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
+# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
+# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
+# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
+# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
+# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
+# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
+# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
+# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
+# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
+# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
+# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
+# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
+# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
+# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
+# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
+# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
+# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
+# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
+# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
+# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
+# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
+# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
+# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
+# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
+# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
+# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
+# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
+# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
+# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
+# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
+# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
+# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
+# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
+# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
+# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
+# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
+# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
+# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
+# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
+# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
+# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
+# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
+# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
+# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
+# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
+# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
+# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
+# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
+# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
+# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
+# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
+# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
+# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
+# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
+# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
+# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
+# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
+# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
+# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
+# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
+# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
+# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
+# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
+# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
+# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
+# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
+# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
+# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
+# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
+# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
+# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
+# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
+# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
+# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
+# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
+# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
+# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
+# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
+# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
+# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
+# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
+# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
+# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
+# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
+# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
+# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
+# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
+# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
+# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
+# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
+# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
+# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
+# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
+# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
+# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
+# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
+# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
+# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
+# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
+# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
+# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
+# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
+# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
+# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
+# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
+# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
+# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
+# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
+# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
+# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
+# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
+# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
+# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
+# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
+# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
+# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
+# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
+# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
+# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
+# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
+# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
+# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
+# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
+# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
+# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
+# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
+# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
+# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
+# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
+# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
+# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
+# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
+# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
+# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
+# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
+# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
+# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
+# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
+# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
+# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
+# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
+# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
+# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
+# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
+# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
+# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
+# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
+# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
+# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
+# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
+# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
+# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
+# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
+# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
+# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
+# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
+# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
+# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
+# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
+# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
+# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
+# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
+# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
+# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
+# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
+# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
+# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
+# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
+# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
+# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
+# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
+# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
+# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
+# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
+# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
+# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
+# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
+# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
+# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
+# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
+# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
+# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
+# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
+# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
+# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
+# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
+# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
+# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
+# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
+# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
+# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
+# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
+# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
+# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
+# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
+# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
+# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
+# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
+# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
+# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
+# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
+# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
+# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
+# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
+# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
+# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
+# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
+# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
+# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
+# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
+# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
+# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
+# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
+# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
+# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
+# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
+# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
+# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
+# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
+# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
+# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
+# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
+# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
+# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
+# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
+# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
+# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
+# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
+# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
+# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
+# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
+# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
+# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
+# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
+# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
+# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
+# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
+# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
+# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
+# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
+# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
+# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
+# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
+# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
+# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
+# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
+# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
+# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
+# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
+# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
+# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
+# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
+# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
+# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
+# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
+# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
+# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
+# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
+# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
+# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
+# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
+# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
+# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
+# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
+# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
+# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
+# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
+# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
+# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
+# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
+# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
+# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
+# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
+# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
+# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
+# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
+# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
+# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
+# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
+# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
+# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
+# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
+# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
+# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
+# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
+# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
+# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
+# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
+# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
+# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
+# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
+# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
+# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
+# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
+# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
+# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
+# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
+# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
+# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
+# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
+# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
+# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
+# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
+# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
+# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
+# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
+# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
+# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
+# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
+# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
+# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
+# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
+# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
+# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
+# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
+# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
+# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
+# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
+# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
+# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
+# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
+# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
+# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
+# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
+# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
+# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
+# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
+# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
+# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
+# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
+# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
+# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
+# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
+# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
+# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
+# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
+# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
+# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
+# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
+# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
+# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
+# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
+# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
+# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
+# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
+# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
+# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
+# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
+# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
+# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
+# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
+# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
+# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
+# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
+# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
+# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
+# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
+# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
+# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
+# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
+# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
+# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
+# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
+# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
+# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
+# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
+# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
+# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
+# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
+# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
+# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
+# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
+# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
+# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
+# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
+# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
+# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
+# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
+# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
+# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
+# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
+# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
+# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
+# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
+# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
+# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
+# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
+# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
+# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
+# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
+# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
+# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
+# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
+# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
+# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
+# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
+# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
+# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
+# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
+# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
+# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
+# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
+# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
+# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
+# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
+# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
+# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
+# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
+# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
+# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
+# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
+# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
+# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
+# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
+# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
+# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
+# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
+# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
+# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
+# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
+# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
+# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
+# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
+# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
+# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
+# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
+# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
+# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
+# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
+# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
+# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
+# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
+# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
+# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
+# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
+# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
+# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
+# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
+# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
+# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
+# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
+# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
+# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
+# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
+# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
+# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
+# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
+# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
+# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
+# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
+# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
+# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
+# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
+# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
+# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
+# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
+# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
+# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
+# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
+# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
+# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
+# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
+# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
+# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
+# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
+# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
+# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
+# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
+# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
+# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
+# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
+# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
+# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
+# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
+# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
+# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
+# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
+# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
+# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
+# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
+# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
+# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
+# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
+# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
+# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
+# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
+# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
+# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
+# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
+# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
+# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
+# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
+# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
+# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
+# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
+# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
+# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
+# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
+# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
+# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
+# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
+# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
+# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
+# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
+# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
+# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
+# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
+# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
+# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
+# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
+# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
+# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
+# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
+# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
+# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
+# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
+# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
+# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
+# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
+# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
+# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
+# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
+# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
+# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
+# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
+# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
+# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
+# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
+# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
+# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
+# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
+# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
+# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
+# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
+# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
+# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
+# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
+# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
+# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
+# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
+# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
+# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
+# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
+# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
+# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
+# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
+# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
+# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
+# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
+# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
+# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
+# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
+# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
+# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
+# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
+# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
+# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
+# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
+# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
+# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
+# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
+# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
+# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
+# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
+# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
+# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
+# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
+# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
+# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
+# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
+# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
+# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
+# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
+# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
+# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
+# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
+# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
+# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
+# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
+# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
+# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
+# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
+# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
+# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
+# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
+# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
+# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
+# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
+# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
+# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
+# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
+# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
+# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
+# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
+# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
+# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
+# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
+# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
+# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
+# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
+# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
+# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
+# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
+# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
+# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
+# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
+# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
+# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
+# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
+# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
+# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
+# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
+# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
+# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
+# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
+# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
+# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
+# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
+# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
+# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
+# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
+# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
+# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
+# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
+# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
+# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
+# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
+# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
+# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
+# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
+# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
+# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
+# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
+# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
+# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
+# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
+# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
+# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
+# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
+# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
+# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
+# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
+# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
+# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
+# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
+# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
+# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
+# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
+# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
+# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
+# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
+# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
+# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
+# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
+# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
+# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
+# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
+# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
+# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
+# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
+# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
+# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
+# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
+# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
+# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
+# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
+# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
+# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
+# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
+# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
+# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
+# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
+# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
+# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
+# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
+# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
+# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
+# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
+# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
+# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
+# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
+# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
+# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
+# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
+# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
+# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
+# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
+# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
+# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
+# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
+# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
+# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
+# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
+# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
+# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
+# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
+# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
+# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
+# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
+# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
+# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
+# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
+# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
+# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
+# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
+# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
+# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
+# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
+# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
+# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
+# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
+# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
+# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
+# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
+# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
+# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
+# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
+# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
+# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
+# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
+# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
+# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
+# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
+# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
+# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
+# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
+# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
+# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
+# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
+# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
+# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
+# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
+# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
+# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
+# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
+# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
+# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
+# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
+# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
+# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
+# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
+# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
+# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
+# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
+# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
+# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
+# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
+# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
+# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
+# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
+# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
+# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
+# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
+# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
+# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
+# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
+# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
+# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
+# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
+# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
+# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
+# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
+# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
+# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
+# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
+# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
+# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
+# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
+# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
+# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
+# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
+# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
+# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
+# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
+# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
+# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
+# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
+# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
+# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
+# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
+# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
+# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
+# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
+# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
+# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
+# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
+# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
+# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
+# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
+# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
+# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
+# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
+# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
+# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
+# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
+# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
+# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
+# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
+# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
+# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
+# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
+# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
+# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
+# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
+# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
+# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
+# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
+# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
+# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
+# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
+# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
+# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
+# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
+# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
+# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
+# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
+# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
+# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
+# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
+# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
+# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
+# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
+# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
+# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
+# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
+# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
+# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
+# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
+# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
+# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
+# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
+# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
+# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
+# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
+# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
+# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
+# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
+# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
+# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
+# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
+# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
+# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
+# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
+# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
+# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
+# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
+# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
+# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
+# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
+# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
+# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
+# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
+# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
+# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
+# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
+# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
+# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
+# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
+# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
+# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
+# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
+# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
+# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
+# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
+# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
+# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
+# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
+# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
+# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
+# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
+# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
+# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
+# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
+# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
+# 11:38:37 --- INFO: Training process end.
+# 11:38:37 --- INFO: Evaluating process start.
+# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
+# 11:38:37 --- INFO: Evaluating process end.
+# 11:38:37 --- INFO: Predicting process start.
+# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
+# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
+# 11:38:38 --- INFO: Predicting process end.
+# 11:38:38 --- INFO: --------------------- GNN end ----------------------
+# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
+# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
+# 11:38:38 --- INFO: Loading dataset.
+# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:38:39 --- INFO: Calculating NTScore for each niche.
+# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
+# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
+# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
+# 11:38:39 --- INFO: Output NTScore tables.
+# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
16.1.3 visualization
16.1.3.1 load ONTraC results
-
+
The NTScore and binarized niche cluster info were stored in cell metadata
-
-# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
-# <char> <char> <char> <char> <char> <char> <char> <int> <num>
-# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
-# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
-# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
-# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
-# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
-# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
+
+# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
+# <char> <char> <char> <char> <char> <char> <char> <int> <num>
+# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
+# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
+# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
+# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
+# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
+# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
The probability matrix of each cell assigned to each niche cluster and
connectivity between niche cluster were stored here.
-
-
+
+
16.1.3.2 niche cluster prob
-spatFeatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'niche cluster',
- expression_values = "prob",
- group_by = 'list_ID',
- feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
- point_border_col = 'gray'
- )
+spatFeatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'niche cluster',
+ expression_values = "prob",
+ group_by = 'list_ID',
+ feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
+ point_border_col = 'gray'
+ )
16.1.3.3 binarized niche cluster
-spatPlot2D(giotto_obj,
- spat_unit = "cell",
- group_by = 'list_ID',
- cell_color = "NicheCluster",
- color_as_factor = T,
- point_size = 1,
- point_border_stroke = NA,
- title="Niche cluster")
+spatPlot2D(giotto_obj,
+ spat_unit = "cell",
+ group_by = 'list_ID',
+ cell_color = "NicheCluster",
+ color_as_factor = T,
+ point_size = 1,
+ point_border_stroke = NA,
+ title="Niche cluster")
16.1.3.5 NT score
-spatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'rna',
- group_by = 'list_ID',
- cell_color = "NTScore",
- color_as_factor = FALSE,
- cell_color_gradient = "turbo",
- point_size = 1,
- point_border_stroke = NA
- )
+spatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ group_by = 'list_ID',
+ cell_color = "NTScore",
+ color_as_factor = FALSE,
+ cell_color_gradient = "turbo",
+ point_size = 1,
+ point_border_stroke = NA
+ )
-
+
diff --git a/interoperability-with-other-frameworks.html b/interoperability-with-other-frameworks.html
index 7050533..347d3af 100644
--- a/interoperability-with-other-frameworks.html
+++ b/interoperability-with-other-frameworks.html
@@ -521,10 +521,10 @@ 15 Interoperability with other fr
15.1 Load Giotto object
To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process:
-
+
Load a Giotto mini Visium object, which will be used for demonstrating interoperability.
-
+
15.2 Seurat
@@ -533,99 +533,99 @@ 15.2 Seurat
15.2.1 Conversion of Giotto Object to Seurat Object
To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function
-
+
Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data.
-gToS <- giottoToSeuratV5(gobject = gobject,
- spat_unit = "cell")
-
-plot1 <- VlnPlot(gToS,
- features = "nCount_rna",
- pt.size = 0.1) +
- NoLegend()
-
-plot2 <- SpatialFeaturePlot(gToS,
- features = "nCount_rna",
- pt.size.factor = 2) +
- theme(legend.position = "right")
-
-wrap_plots(plot1, plot2)
+gToS <- giottoToSeuratV5(gobject = gobject,
+ spat_unit = "cell")
+
+plot1 <- VlnPlot(gToS,
+ features = "nCount_rna",
+ pt.size = 0.1) +
+ NoLegend()
+
+plot2 <- SpatialFeaturePlot(gToS,
+ features = "nCount_rna",
+ pt.size.factor = 2) +
+ theme(legend.position = "right")
+
+wrap_plots(plot1, plot2)
15.2.1.1 Apply SCTransform
We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function.
-
+
15.2.1.2 Dimension Reduction
We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object:
-
+
15.2.2 Conversion of Seurat object Back to Giotto Object
To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks.
-giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
- spatial_assay = "rna",
- dim_reduction = c("pca", "umap"),
- sp_network = "Delaunay_network",
- nn_network = c("sNN.pca", "custom_NN" ))
+giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
+ spatial_assay = "rna",
+ dim_reduction = c("pca", "umap"),
+ sp_network = "Delaunay_network",
+ nn_network = c("sNN.pca", "custom_NN" ))
15.2.2.1 Clustering and Plotting UMAP
Here we perform K-means clustering on the UMAP results obtained from the Seurat object:
-## k-means clustering
-giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
- dim_reduction_to_use = "pca")
-
-#Plot UMAP post-clustering to visualize kmeans
-graph2 <- Giotto::plotUMAP(
- gobject = giottoFromSeurat,
- cell_color = "kmeans",
- show_NN_network = TRUE,
- point_size = 2.5
-)
+## k-means clustering
+giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
+ dim_reduction_to_use = "pca")
+
+#Plot UMAP post-clustering to visualize kmeans
+graph2 <- Giotto::plotUMAP(
+ gobject = giottoFromSeurat,
+ cell_color = "kmeans",
+ show_NN_network = TRUE,
+ point_size = 2.5
+)
15.2.2.2 Spatial CoExpression
We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function:
-ranktest <- binSpect(giottoFromSeurat,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "Delaunay_network")
-
-ext_spatial_genes <- ranktest[1:300,]$feats
-
-spat_cor_netw_DT <- detectSpatialCorFeats(
- giottoFromSeurat,
- method = "network",
- spatial_network_name = "Delaunay_network",
- subset_feats = ext_spatial_genes)
-
-top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
- feats = "Dsp",
- show_top_feats = 10)
-
-spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
- name = "spat_netw_clus",
- k = 7)
-heatmSpatialCorFeats(
- giottoFromSeurat,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL),
- save_plot = TRUE,
- show_plot = TRUE,
- return_plot = FALSE,
- save_param = list(base_height = 6, base_width = 8, units = 'cm'))
+ranktest <- binSpect(giottoFromSeurat,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "Delaunay_network")
+
+ext_spatial_genes <- ranktest[1:300,]$feats
+
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ giottoFromSeurat,
+ method = "network",
+ spatial_network_name = "Delaunay_network",
+ subset_feats = ext_spatial_genes)
+
+top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
+ feats = "Dsp",
+ show_top_feats = 10)
+
+spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
+ name = "spat_netw_clus",
+ k = 7)
+heatmSpatialCorFeats(
+ giottoFromSeurat,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ return_plot = FALSE,
+ save_param = list(base_height = 6, base_width = 8, units = 'cm'))
@@ -635,60 +635,60 @@ 15.3 SpatialExperiment
To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries.
-library(SpatialExperiment)
-library(ggspavis)
-library(pheatmap)
-library(scater)
-library(scran)
-library(nnSVG)
+library(SpatialExperiment)
+library(ggspavis)
+library(pheatmap)
+library(scater)
+library(scran)
+library(nnSVG)
15.3.1 Convert Giotto Object to SpatialExperiment Object
To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function.
-
+
The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use:
-
+
15.3.1.1 Identify top spatially variable genes with nnSVG
We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate:
-# One of the assays should be "logcounts"
-# We rename the normalized assay to "logcounts"
-assayNames(spe)[[2]] <- "logcounts"
-
-# Create model matrix for leiden clustering labels
-X <- model.matrix(~ colData(spe)$leiden_clus)
-dim(X)
+# One of the assays should be "logcounts"
+# We rename the normalized assay to "logcounts"
+assayNames(spe)[[2]] <- "logcounts"
+
+# Create model matrix for leiden clustering labels
+X <- model.matrix(~ colData(spe)$leiden_clus)
+dim(X)
- Run nnSVG
This step will take several minutes to run
-
+
15.3.2 Conversion of SpatialExperiment object back to Giotto
We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object.
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-print(giottoFromSPE)
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+print(giottoFromSPE)
15.3.2.1 Plotting top genes from nnSVG in Giotto
Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object.
-ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
- output = "data.table")
-
-ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
-spatFeatPlot2D(giottoFromSPE,
- expression_values = "scaled_rna_cell",
- feats = ext_spatial_genes[1:4],
- point_size = 2)
+ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
+ output = "data.table")
+
+ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
+spatFeatPlot2D(giottoFromSPE,
+ expression_values = "scaled_rna_cell",
+ feats = ext_spatial_genes[1:4],
+ point_size = 2)
@@ -701,97 +701,97 @@ 15.4 AnnData
15.4.1 Load Required Libraries
To start, we need to load the necessary libraries, including reticulate for interfacing with Python.
-
+
15.4.2 Specify Path for Results
First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions.
-# Specify path to which results may be saved
-results_directory <- "results/03_session4/giotto_anndata_conversion/"
-
-instrs <- showGiottoInstructions(gobject)
-
-mini_gobject <- replaceGiottoInstructions(gobject = gobject,
- instructions = instrs)
+# Specify path to which results may be saved
+results_directory <- "results/03_session4/giotto_anndata_conversion/"
+
+instrs <- showGiottoInstructions(gobject)
+
+mini_gobject <- replaceGiottoInstructions(gobject = gobject,
+ instructions = instrs)
15.4.2.1 Create Default kNN Network
We will create a k-nearest neighbor (kNN) network using mostly default parameters.
-
+
15.4.3 Giotto To AnnData
To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData()
-
+
Next, we import scanpy and perform a series of preprocessing steps on the AnnData object.
-scanpy <- import("scanpy")
-
-adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
-
-# Normalize total counts per cell
-scanpy$pp$normalize_total(adata, target_sum=1e4)
-
-# Log-transform the data
-scanpy$pp$log1p(adata)
-
-# Perform PCA
-scanpy$pp$pca(adata, n_comps=40L)
-
-# Compute the neighborhood graph
-scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
-
-# Run UMAP
-scanpy$tl$umap(adata)
-
-# Save the processed AnnData object
-adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
-
-processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
+scanpy <- import("scanpy")
+
+adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
+
+# Normalize total counts per cell
+scanpy$pp$normalize_total(adata, target_sum=1e4)
+
+# Log-transform the data
+scanpy$pp$log1p(adata)
+
+# Perform PCA
+scanpy$pp$pca(adata, n_comps=40L)
+
+# Compute the neighborhood graph
+scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
+
+# Run UMAP
+scanpy$tl$umap(adata)
+
+# Save the processed AnnData object
+adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
+
+processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
15.4.4 Convert AnnData to Giotto
Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto.
-
+
15.4.4.1 UMAP Visualization
Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object.
-Giotto::plotUMAP(
- gobject = giottoFromAnndata,
- dim_reduction_name = "umap.ad",
- cell_color = "leiden_clus",
- point_size = 3
-)
+Giotto::plotUMAP(
+ gobject = giottoFromAnndata,
+ dim_reduction_name = "umap.ad",
+ cell_color = "leiden_clus",
+ point_size = 3
+)
15.5 Create mini Vizgen object
-mini_gobject <- loadGiottoMini(dataset = "vizgen",
- python_path = NULL)
-
-mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
- instructions = instrs)
-mini_gobject <- createNearestNetwork(gobject = mini_gobject,
- spat_unit = "aggregate",
- feat_type = "rna",
- type = "kNN",
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap",
- k = 6,
- name = "new_network")
+mini_gobject <- loadGiottoMini(dataset = "vizgen",
+ python_path = NULL)
+
+mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
+ instructions = instrs)
+mini_gobject <- createNearestNetwork(gobject = mini_gobject,
+ spat_unit = "aggregate",
+ feat_type = "rna",
+ type = "kNN",
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap",
+ k = 6,
+ name = "new_network")
Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details.
-
+
diff --git a/introduction-to-the-giotto-package.html b/introduction-to-the-giotto-package.html
index e22d6d1..0ab2014 100644
--- a/introduction-to-the-giotto-package.html
+++ b/introduction-to-the-giotto-package.html
@@ -546,19 +546,58 @@ 4.3 Installation + python environ
4.3.1 Giotto installation
Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository.
Much of this already covered in Section 2.2, but the highlights are:
-
-- Ensure your system
-
-To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak.
+
+4.3.1.1 System prerequisites
+
+- for windows, Rtools needs to be installed
+- a major dependency terra needs
GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux
+
+
+
+4.3.1.2 Installation of released version
+To install the currently released version of Giotto in a single step:
+This should automatically install all the Giotto dependencies and other Giotto module packages.
+
+
+4.3.1.3 Installation of dev branch Giotto packages
+You can install dev branch versions by using devtools::install_github()
+Core module dev branchs:
+
+- “drieslab/Giotto@suite_dev”
+- “drieslab/GiottoVisuals@dev”
+- “drieslab/GiottoClass@dev”
+- “drieslab/GiottoUtils@dev”
+
+
+pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages.
+
+
+4.3.1.4 Common install issues
+If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work.
+
+If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it.
+
+## # Downloading packages -------------------------------------------------------
+## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s]
+## Successfully downloaded 1 package in 1.2 seconds.
+##
+## The following package(s) will be installed:
+## - irlba [2.3.5.1]
+## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu".
+##
+## # Installing packages --------------------------------------------------------
+## - Installing irlba ... OK [built from source and cached in 5.7s]
+## Successfully installed 1 package in 5.7 seconds.
+
4.3.2 Python environment
4.3.2.1 Default installation
In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment.
-
+
Two things are needed for python to work:
- A conda (e.g. miniconda or anaconda) installation which is the package and environment management system.
@@ -580,17 +619,17 @@ 4.3.2.1 Default installation
4.3.2.2 Custom installs
Custom python environments can be made by first setting up a new environment and establishing the name and python version to use.
-
+
Following that, one or more python packages to install can be added to the environment.
-reticulate::py_install(
- pip = TRUE,
- envname = '[name of env]',
- packages = c(
- "package1",
- "package2",
- "..."
- )
-)
+reticulate::py_install(
+ pip = TRUE,
+ envname = '[name of env]',
+ packages = c(
+ "package1",
+ "package2",
+ "..."
+ )
+)
Once an environment has been set up, Giotto can hook into it.
@@ -612,17 +651,56 @@ 4.3.2.3 Using a specific environm
Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto.
You would run file.edit("~/.Rprofile") and then add options("giotto.py_path" = "[path to env or envname]") as a line so that it is automatically set at the start of each session.
If a specific environment should only be used a couple times then method 1 is easiest:
-
+
To check which conda environments exist on your machine:
-
+
Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running:
-
+
4.4 Giotto instructions
-Giotto
+Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are:
+
+- python_path - when set, will activate a python environment
+- save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render.
+- save_plot - whether to save plots to the
save_dir
+- return_plot - whether to return the plot objects. When FALSE, only NULL is returned
+- show_plot - whether to show the plot in the viewer
+
+These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function.
+library(Giotto)
+save_dir <- "results/01_session2/"
+
+# this call will also intialize the python env
+instrs <- createGiottoInstructions(
+ save_dir = save_dir, # working directory is the default
+ show_plot = FALSE,
+ save_plot = TRUE,
+ return_plot = FALSE,
+ python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default
+)
+force(instrs)
+Giotto object creation functions all have an instructions param for passing in instructions objects.
+giotto objects will also respond to the instructions() generic.
+test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object
+
+# example plot
+g <- GiottoData::loadGiottoMini("visium")
+instructions(g) <- instrs
+instructions(g, "show_plot") # instructions say not to plot to viewer
+spatPlot2D(g, show_image = TRUE, image_name = "image")
+# instead it will directly write to the results folder
+As an example, you can also set individual instructions
+
+
+
+
+Figure 4.2: example image output
+
+
diff --git a/multi-omics-integration.html b/multi-omics-integration.html
index 47cad23..e76e25a 100644
--- a/multi-omics-integration.html
+++ b/multi-omics-integration.html
@@ -520,14 +520,14 @@ 14 Multi-omics integration
14.1 The CytAssist technology
The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image.
-
+
Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics.
The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument.
-
+
Figure 14.2: CytAssist workflow. Source: 10X genomics.
@@ -542,19 +542,19 @@
14.2 Introduction to the spatial
14.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/03_session3")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
+dir.create("data/03_session3")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”.
-
+
14.4 Create the Giotto object
@@ -564,124 +564,124 @@ 14.4 Create the Giotto objectx,y(,z) coordinates for cells or spots (or the path to)
createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/03_session3/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-# Provide the path to the data folder
-data_path <- "data/03_session3/"
-
-# Create object directly from the data folder
-visium_tonsil <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/03_session3/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+# Provide the path to the data folder
+data_path <- "data/03_session3/"
+
+# Create object directly from the data folder
+visium_tonsil <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
- Print the information of the object, note that both rna and protein are listed in the expression slot.
-
+
14.5 Subset on spots that were covered by tissue
-spatPlot2D(
- gobject = visium_tonsil,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE,
- image_name = "image"
-)
-
+spatPlot2D(
+ gobject = visium_tonsil,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE,
+ image_name = "image"
+)
+
Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object.
-
+
14.6 RNA processing
- Run the Filtering, normalization, and statistics steps using only the RNA feature.
-visium_tonsil <- filterGiotto(
- gobject = visium_tonsil,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil)
+visium_tonsil <- filterGiotto(
+ gobject = visium_tonsil,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1000,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil)
- Dimension reduction
Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs.
-visium_tonsil <- calculateHVF(gobject = visium_tonsil)
-
-visium_tonsil <- runPCA(gobject = visium_tonsil)
+visium_tonsil <- calculateHVF(gobject = visium_tonsil)
+
+visium_tonsil <- runPCA(gobject = visium_tonsil)
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the RNA-based Leiden clusters.
-
+
- Visualization
Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot.
-plotUMAP(gobject = visium_tonsil,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters.
Plot the spatial distribution of the RNA-based Leiden cluster per spot.
-spatPlot2D(gobject = visium_tonsil,
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.5: Spatial distribution of RNA-based Leiden clusters.
@@ -693,80 +693,80 @@
14.7 Protein processingvisium_tonsil <- filterGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein")
+visium_tonsil <- filterGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein")
- Dimension reduction
Calculate the principal components using all the proteins available in the dataset.
-
+
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the Protein-based Leiden clusters.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- resolution = 1,
- n_iterations = 1000)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ resolution = 1,
+ n_iterations = 1000)
- Visualization
Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters.
Plot the spatial distribution of the Protein-based Leiden clusters.
-spatPlot2D(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.7: Spatial distribution of Protein-based Leiden clusters.
@@ -779,68 +779,68 @@
14.8 Multi-omics integrationCalculate kNN
Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type.
-## RNA modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
-
-## Protein modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
+## RNA modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
+
+## Protein modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
- Run WNN
Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object.
-visium_tonsil <- runWNN(visium_tonsil,
- spat_unit = "cell",
- modality_1 = "rna",
- modality_2 = "protein",
- pca_name_modality_1 = "pca",
- pca_name_modality_2 = "protein.pca",
- k = 20,
- integrated_feat_type = NULL,
- matrix_result_name = NULL,
- w_name_modality_1 = NULL,
- w_name_modality_2 = NULL,
- verbose = TRUE)
+visium_tonsil <- runWNN(visium_tonsil,
+ spat_unit = "cell",
+ modality_1 = "rna",
+ modality_2 = "protein",
+ pca_name_modality_1 = "pca",
+ pca_name_modality_2 = "protein.pca",
+ k = 20,
+ integrated_feat_type = NULL,
+ matrix_result_name = NULL,
+ w_name_modality_1 = NULL,
+ w_name_modality_2 = NULL,
+ verbose = TRUE)
- Run Integrated umap
Calculate the UMAP using the weights of each feature per spot.
-visium_tonsil <- runIntegratedUMAP(visium_tonsil,
- modality1 = "rna",
- modality2 = "protein",
- spread = 7,
- min_dist = 1,
- force = FALSE)
+visium_tonsil <- runIntegratedUMAP(visium_tonsil,
+ modality1 = "rna",
+ modality2 = "protein",
+ spread = 7,
+ min_dist = 1,
+ force = FALSE)
- Calculate integrated clusters
Calculate the multiomics-based Leiden clusters using the weights of each feature per spot.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- nn_network_to_use = "kNN",
- network_name = "integrated_kNN",
- name = "integrated_leiden_clus",
- resolution = 1)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ nn_network_to_use = "kNN",
+ network_name = "integrated_kNN",
+ name = "integrated_leiden_clus",
+ resolution = 1)
- Visualize the integrated UMAP
Plot the integrated UMAP and color the spots using the integrated Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- dim_reduction_name = "integrated.umap",
- point_size = 2,
- title = "Integrated UMAP using Integrated Leiden clusters")
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ dim_reduction_name = "integrated.umap",
+ point_size = 2,
+ title = "Integrated UMAP using Integrated Leiden clusters")
+
Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters.
@@ -850,14 +850,14 @@
14.8 Multi-omics integrationVisualize spatial plot with integrated clusters
Plot the spatial distribution of the integrated Leiden clusters.
-spatPlot2D(visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- point_size = 3,
- show_image = TRUE,
- title = "Integrated Leiden clustering")
-
+spatPlot2D(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ point_size = 3,
+ show_image = TRUE,
+ title = "Integrated Leiden clustering")
+
Figure 14.9: Spatial distribution of the integrated Leiden clusters.
@@ -866,85 +866,85 @@
14.8 Multi-omics integration
14.9 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
- [7] matrixStats_1.3.0 compiler_4.4.1
- [9] png_0.1-8 systemfonts_1.1.0
- [11] vctrs_0.6.5 reshape2_1.4.4
- [13] stringr_1.5.1 pkgconfig_2.0.3
- [15] SpatialExperiment_1.14.0 crayon_1.5.3
- [17] fastmap_1.2.0 backports_1.5.0
- [19] magick_2.8.4 XVector_0.44.0
- [21] labeling_0.4.3 utf8_1.2.4
- [23] rmarkdown_2.27 UCSC.utils_1.0.0
- [25] ragg_1.3.2 purrr_1.0.2
- [27] xfun_0.46 beachmat_2.20.0
- [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
- [31] jsonlite_1.8.8 DelayedArray_0.30.1
- [33] BiocParallel_1.38.0 terra_1.7-78
- [35] irlba_2.3.5.1 parallel_4.4.1
- [37] R6_2.5.1 stringi_1.8.4
- [39] RColorBrewer_1.1-3 reticulate_1.38.0
- [41] parallelly_1.37.1 GenomicRanges_1.56.1
- [43] scattermore_1.2 Rcpp_1.0.13
- [45] bookdown_0.40 SummarizedExperiment_1.34.0
- [47] knitr_1.48 future.apply_1.11.2
- [49] R.utils_2.12.3 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
+ [7] matrixStats_1.3.0 compiler_4.4.1
+ [9] png_0.1-8 systemfonts_1.1.0
+ [11] vctrs_0.6.5 reshape2_1.4.4
+ [13] stringr_1.5.1 pkgconfig_2.0.3
+ [15] SpatialExperiment_1.14.0 crayon_1.5.3
+ [17] fastmap_1.2.0 backports_1.5.0
+ [19] magick_2.8.4 XVector_0.44.0
+ [21] labeling_0.4.3 utf8_1.2.4
+ [23] rmarkdown_2.27 UCSC.utils_1.0.0
+ [25] ragg_1.3.2 purrr_1.0.2
+ [27] xfun_0.46 beachmat_2.20.0
+ [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
+ [31] jsonlite_1.8.8 DelayedArray_0.30.1
+ [33] BiocParallel_1.38.0 terra_1.7-78
+ [35] irlba_2.3.5.1 parallel_4.4.1
+ [37] R6_2.5.1 stringi_1.8.4
+ [39] RColorBrewer_1.1-3 reticulate_1.38.0
+ [41] parallelly_1.37.1 GenomicRanges_1.56.1
+ [43] scattermore_1.2 Rcpp_1.0.13
+ [45] bookdown_0.40 SummarizedExperiment_1.34.0
+ [47] knitr_1.48 future.apply_1.11.2
+ [49] R.utils_2.12.3 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/reference-keys.txt b/reference-keys.txt
index a17604d..555f956 100644
--- a/reference-keys.txt
+++ b/reference-keys.txt
@@ -4,115 +4,116 @@ fig:unnamed-chunk-14
fig:unnamed-chunk-15
fig:unnamed-chunk-16
fig:unnamed-chunk-18
-fig:unnamed-chunk-30
-fig:unnamed-chunk-35
-fig:unnamed-chunk-38
-fig:unnamed-chunk-40
+fig:unnamed-chunk-32
+fig:unnamed-chunk-37
fig:unnamed-chunk-42
-fig:unnamed-chunk-44
+fig:unnamed-chunk-45
+fig:unnamed-chunk-47
fig:unnamed-chunk-49
fig:unnamed-chunk-51
-fig:unnamed-chunk-53
-fig:unnamed-chunk-55
-fig:unnamed-chunk-59
-fig:unnamed-chunk-61
-fig:unnamed-chunk-63
+fig:unnamed-chunk-56
+fig:unnamed-chunk-58
+fig:unnamed-chunk-60
+fig:unnamed-chunk-62
fig:unnamed-chunk-66
-fig:unnamed-chunk-69
-fig:unnamed-chunk-74
+fig:unnamed-chunk-68
+fig:unnamed-chunk-70
+fig:unnamed-chunk-73
fig:unnamed-chunk-76
-fig:unnamed-chunk-78
-fig:unnamed-chunk-80
-fig:unnamed-chunk-82
-fig:unnamed-chunk-84
+fig:unnamed-chunk-81
+fig:unnamed-chunk-83
+fig:unnamed-chunk-85
fig:unnamed-chunk-87
+fig:unnamed-chunk-89
+fig:unnamed-chunk-91
fig:unnamed-chunk-94
-fig:unnamed-chunk-96
-fig:unnamed-chunk-98
fig:unnamed-chunk-101
fig:unnamed-chunk-103
fig:unnamed-chunk-105
+fig:unnamed-chunk-108
+fig:unnamed-chunk-110
fig:unnamed-chunk-112
-fig:unnamed-chunk-114
-fig:unnamed-chunk-118
-fig:unnamed-chunk-120
-fig:unnamed-chunk-123
-fig:unnamed-chunk-128
-fig:unnamed-chunk-131
-fig:unnamed-chunk-134
-fig:unnamed-chunk-137
+fig:unnamed-chunk-119
+fig:unnamed-chunk-121
+fig:unnamed-chunk-125
+fig:unnamed-chunk-127
+fig:unnamed-chunk-130
+fig:unnamed-chunk-135
+fig:unnamed-chunk-138
fig:unnamed-chunk-141
-fig:unnamed-chunk-143
-fig:unnamed-chunk-147
+fig:unnamed-chunk-144
+fig:unnamed-chunk-148
fig:unnamed-chunk-150
-fig:unnamed-chunk-152
+fig:unnamed-chunk-154
+fig:unnamed-chunk-157
fig:unnamed-chunk-159
-fig:unnamed-chunk-161
-fig:unnamed-chunk-163
-fig:unnamed-chunk-165
-fig:unnamed-chunk-167
-fig:unnamed-chunk-169
-fig:unnamed-chunk-171
+fig:unnamed-chunk-166
+fig:unnamed-chunk-168
+fig:unnamed-chunk-170
+fig:unnamed-chunk-172
fig:unnamed-chunk-174
-fig:unnamed-chunk-175
fig:unnamed-chunk-176
-fig:unnamed-chunk-185
-fig:unnamed-chunk-191
-fig:unnamed-chunk-194
+fig:unnamed-chunk-178
+fig:unnamed-chunk-181
+fig:unnamed-chunk-182
+fig:unnamed-chunk-183
+fig:unnamed-chunk-192
+fig:unnamed-chunk-198
fig:unnamed-chunk-201
-fig:unnamed-chunk-204
-fig:unnamed-chunk-206
-fig:unnamed-chunk-210
-fig:unnamed-chunk-212
-fig:unnamed-chunk-215
+fig:unnamed-chunk-208
+fig:unnamed-chunk-211
+fig:unnamed-chunk-213
fig:unnamed-chunk-217
fig:unnamed-chunk-219
-fig:unnamed-chunk-221
+fig:unnamed-chunk-222
fig:unnamed-chunk-224
-fig:unnamed-chunk-225
-fig:unnamed-chunk-227
-fig:unnamed-chunk-229
-fig:unnamed-chunk-230
+fig:unnamed-chunk-226
+fig:unnamed-chunk-228
+fig:unnamed-chunk-231
fig:unnamed-chunk-232
fig:unnamed-chunk-234
fig:unnamed-chunk-236
-fig:unnamed-chunk-275
-fig:unnamed-chunk-279
-fig:unnamed-chunk-281
-fig:unnamed-chunk-283
+fig:unnamed-chunk-237
+fig:unnamed-chunk-239
+fig:unnamed-chunk-241
+fig:unnamed-chunk-243
+fig:unnamed-chunk-282
fig:unnamed-chunk-286
fig:unnamed-chunk-288
fig:unnamed-chunk-290
-fig:unnamed-chunk-292
-fig:unnamed-chunk-294
+fig:unnamed-chunk-293
+fig:unnamed-chunk-295
fig:unnamed-chunk-297
fig:unnamed-chunk-299
fig:unnamed-chunk-301
-fig:unnamed-chunk-303
-fig:unnamed-chunk-305
-fig:unnamed-chunk-365
-fig:unnamed-chunk-366
+fig:unnamed-chunk-304
+fig:unnamed-chunk-306
+fig:unnamed-chunk-308
+fig:unnamed-chunk-310
+fig:unnamed-chunk-312
fig:unnamed-chunk-372
+fig:unnamed-chunk-373
fig:unnamed-chunk-379
-fig:unnamed-chunk-381
-fig:unnamed-chunk-387
-fig:unnamed-chunk-389
-fig:unnamed-chunk-395
-fig:unnamed-chunk-397
-fig:unnamed-chunk-462
-fig:unnamed-chunk-465
+fig:unnamed-chunk-386
+fig:unnamed-chunk-388
+fig:unnamed-chunk-394
+fig:unnamed-chunk-396
+fig:unnamed-chunk-402
+fig:unnamed-chunk-404
fig:unnamed-chunk-469
-fig:unnamed-chunk-470
fig:unnamed-chunk-472
-fig:unnamed-chunk-474
fig:unnamed-chunk-476
-fig:unnamed-chunk-478
+fig:unnamed-chunk-477
fig:unnamed-chunk-479
fig:unnamed-chunk-481
+fig:unnamed-chunk-483
fig:unnamed-chunk-485
+fig:unnamed-chunk-486
fig:unnamed-chunk-488
-fig:unnamed-chunk-490
fig:unnamed-chunk-492
+fig:unnamed-chunk-495
+fig:unnamed-chunk-497
+fig:unnamed-chunk-499
giotto-suite-workshop-2024
instructors
topics-and-schedule
@@ -153,6 +154,10 @@ presentation-1
ecosystem
installation-python-environment
giotto-installation-1
+system-prerequisites
+installation-of-released-version
+installation-of-dev-branch-giotto-packages
+common-install-issues
python-environment
default-installation
custom-installs
diff --git a/search_index.json b/search_index.json
index 2736898..72ab7c3 100644
--- a/search_index.json
+++ b/search_index.json
@@ -1 +1 @@
-[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: Ensure your system To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak. pak::pak("drieslab/Giotto") 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
+[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: 4.3.1.1 System prerequisites for windows, Rtools needs to be installed a major dependency terra needs GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux 4.3.1.2 Installation of released version To install the currently released version of Giotto in a single step: pak::pak("drieslab/Giotto") This should automatically install all the Giotto dependencies and other Giotto module packages. 4.3.1.3 Installation of dev branch Giotto packages You can install dev branch versions by using devtools::install_github() Core module dev branchs: “drieslab/Giotto@suite_dev” “drieslab/GiottoVisuals@dev” “drieslab/GiottoClass@dev” “drieslab/GiottoUtils@dev” devtools::install_github("drieslab/GiottoClass@dev") pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages. 4.3.1.4 Common install issues If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work. devtools::install_version("Matrix", version = "1.6-5") If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it. install.packages("irlba", type = "source") ## # Downloading packages ------------------------------------------------------- ## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s] ## Successfully downloaded 1 package in 1.2 seconds. ## ## The following package(s) will be installed: ## - irlba [2.3.5.1] ## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu". ## ## # Installing packages -------------------------------------------------------- ## - Installing irlba ... OK [built from source and cached in 5.7s] ## Successfully installed 1 package in 5.7 seconds. 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are: python_path - when set, will activate a python environment save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render. save_plot - whether to save plots to the save_dir return_plot - whether to return the plot objects. When FALSE, only NULL is returned show_plot - whether to show the plot in the viewer These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function. library(Giotto) save_dir <- "results/01_session2/" # this call will also intialize the python env instrs <- createGiottoInstructions( save_dir = save_dir, # working directory is the default show_plot = FALSE, save_plot = TRUE, return_plot = FALSE, python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default ) force(instrs) Giotto object creation functions all have an instructions param for passing in instructions objects. giotto objects will also respond to the instructions() generic. test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object # example plot g <- GiottoData::loadGiottoMini("visium") instructions(g) <- instrs instructions(g, "show_plot") # instructions say not to plot to viewer spatPlot2D(g, show_image = TRUE, image_name = "image") # instead it will directly write to the results folder As an example, you can also set individual instructions instructions(g, "show_plot") <- TRUE spatPlot2D(g, show_image = TRUE, image_name = "image") Figure 4.2: example image output "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
diff --git a/spatial-multi-modal-analysis.html b/spatial-multi-modal-analysis.html
index 5bdab74..c78f952 100644
--- a/spatial-multi-modal-analysis.html
+++ b/spatial-multi-modal-analysis.html
@@ -563,48 +563,48 @@ 13.2.3 Spatial classes:# load in data
-library(Giotto)
-
-g <- GiottoData::loadGiottoMini("vizgen")
-
-activeSpatUnit(g) <- "aggregate"
-
-gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE)
-
-gimg <- getGiottoImage(g)
+
13.3 Examples of the simple transforms with a giottoPolygon
-rain <- rainbow(nrow(gpoly))
-
-line_width <- 0.3
-
-# par to setup the grid plotting layout
-p <- par(no.readonly = TRUE)
-par(mfrow=c(3,3))
-gpoly |>
- plot(main = "no transform", col = rain, lwd = line_width)
-
-gpoly |> spatShift(dx = 1000) |>
- plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
-
-gpoly |> spin(45) |>
- plot(main = "spin(45)", col = rain, lwd = line_width)
-
-gpoly |> rescale(fx = 10, fy = 6) |>
- plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
-
-gpoly |> flip(direction = "vertical") |>
- plot(main = "flip()", col = rain, lwd = line_width)
-
-gpoly |> t() |>
- plot(main = "t()", col = rain, lwd = line_width)
-
-gpoly |> shear(fx = 0.5) |>
- plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
-par(p)
+rain <- rainbow(nrow(gpoly))
+
+line_width <- 0.3
+
+# par to setup the grid plotting layout
+p <- par(no.readonly = TRUE)
+par(mfrow=c(3,3))
+gpoly |>
+ plot(main = "no transform", col = rain, lwd = line_width)
+
+gpoly |> spatShift(dx = 1000) |>
+ plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
+
+gpoly |> spin(45) |>
+ plot(main = "spin(45)", col = rain, lwd = line_width)
+
+gpoly |> rescale(fx = 10, fy = 6) |>
+ plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
+
+gpoly |> flip(direction = "vertical") |>
+ plot(main = "flip()", col = rain, lwd = line_width)
+
+gpoly |> t() |>
+ plot(main = "t()", col = rain, lwd = line_width)
+
+gpoly |> shear(fx = 0.5) |>
+ plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
+par(p)
@@ -617,20 +617,20 @@ 13.4 Affine transforms# create affine2d
-aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
+# create affine2d
+aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set.
-
-# append several simple transforms
-aff <- aff |>
- spatShift(dx = 1000) |>
- spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- flip(direction = "vertical") |>
- t() |>
- shear(fx = 0.5)
-force(aff)
+
+# append several simple transforms
+aff <- aff |>
+ spatShift(dx = 1000) |>
+ spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ flip(direction = "vertical") |>
+ t() |>
+ shear(fx = 0.5)
+force(aff)
<affine2d>
anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax)
rotate : -0.785398163397448 (rad)
@@ -639,35 +639,35 @@ 13.4 Affine transformsplot(aff)
+
We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests.
-
+
13.5 Image transforms
Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting.
spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image.
-# affine transform of image info matches with polygon info
-gimg |> affine(aff) |> plot()
-gpoly |> affine(aff) |>
- plot(add = TRUE,
- border = "cyan",
- lwd = 0.3)
+# affine transform of image info matches with polygon info
+gimg |> affine(aff) |> plot()
+gpoly |> affine(aff) |>
+ plot(add = TRUE,
+ border = "cyan",
+ lwd = 0.3)
-# affine of the giotto object
-g |> affine(aff) |>
- spatInSituPlotPoints(
- show_image = TRUE,
- feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
- feats_color_code = rainbow(4),
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- point_size = 0.1,
- use_overlap = FALSE
- )
+# affine of the giotto object
+g |> affine(aff) |>
+ spatInSituPlotPoints(
+ show_image = TRUE,
+ feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
+ feats_color_code = rainbow(4),
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ point_size = 0.1,
+ use_overlap = FALSE
+ )
Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used.
@@ -687,256 +687,256 @@ 13.6.1 Example dataset: Xenium Br
– cooresponding centroid locations
The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance.
-library(Giotto)
-instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
- save_plot = TRUE,
- show_plot = TRUE)
-options(timeout = 999999)
-download_dir <-file.path(getwd(),'/data/03_session2/')
-destfile <- file.path(download_dir,'Multimodal_registration.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
-Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
-Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
+library(Giotto)
+instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
+ save_plot = TRUE,
+ show_plot = TRUE)
+options(timeout = 999999)
+download_dir <-file.path(getwd(),'/data/03_session2/')
+destfile <- file.path(download_dir,'Multimodal_registration.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
+Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
+Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
13.6.2 Target Coordinate system
Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system.
-xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
-xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
-xen_cell_pl
+xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
+xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+xen_cell_pl
13.6.3 Visium to register
Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration
-G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
- gene_column_index = 2,
- png_name = 'tissue_hires_image.png',
- instructions = NULL)
-# In the meantime, calculate statistics for easier plot showing
-G_visium <- normalizeGiotto(G_visium)
-G_visium <- addStatistics(G_visium)
-V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
-V_origin
+G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
+ gene_column_index = 2,
+ png_name = 'tissue_hires_image.png',
+ instructions = NULL)
+# In the meantime, calculate statistics for easier plot showing
+G_visium <- normalizeGiotto(G_visium)
+G_visium <- addStatistics(G_visium)
+V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
+V_origin
The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform.
-# create affine2d
-aff <- affine(diag(c(1,1)))
-aff <- aff |>
- spin(90) |>
- flip(direction = "horizontal")
-force(aff)
-
-# Apply the transform
-V_tansformed <- affine(G_visium,aff)
-spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
-spatplot_to_register
+# create affine2d
+aff <- affine(diag(c(1,1)))
+aff <- aff |>
+ spin(90) |>
+ flip(direction = "horizontal")
+force(aff)
+
+# Apply the transform
+V_tansformed <- affine(G_visium,aff)
+spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
+spatplot_to_register
Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images.
Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app.
-
+
Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result.
-landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
-affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
-
-V_final <- affine(G_visium,affine_mtx %*% aff@affine)
-
-spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
-spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
+affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
+
+V_final <- affine(G_visium,affine_mtx %*% aff@affine)
+
+spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
+spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.3.1 Create Pseudo Visium dataset for comparison
Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance.
-V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
-spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
-spatPlot2D(V_final,
- show_network = T,
- network_color = 'blue',
- point_size = 1)
-center_to_center <- min(spat_network$distance)
-radius <- center_to_center*55/200
+V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
+spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
+spatPlot2D(V_final,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1)
+center_to_center <- min(spat_network$distance)
+radius <- center_to_center*55/200
Now we get the Pseudo Visium polygons
-Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
-stamp_dt <- circleVertices(radius = radius, npoints = 100)
-pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
-pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
-plot(pseudo_visium_poly)
+Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
+stamp_dt <- circleVertices(radius = radius, npoints = 100)
+pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
+pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
+plot(pseudo_visium_poly)
Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points
-xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
-gpoints <- createGiottoPoints(xen_transcripts)
-Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
- gpolygons = list('visium' = pseudo_visium_poly))
+xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
+gpoints <- createGiottoPoints(xen_transcripts)
+Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
+ gpolygons = list('visium' = pseudo_visium_poly))
Get gene expression information by overlapping polygon to points
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'visium')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "visium",
- feat_info = "rna",
- aggr_function = "sum")
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'visium')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "visium",
+ feat_info = "rna",
+ aggr_function = "sum")
Manipulate the expression for plotting
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'visium',
- name = 'plot')
-
-spatFeatPlot2D(Xen_obj,
- point_size = 3.5,
- expression_values = 'plot',
- show_image = F,
- feats = 'ERBB2')
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ name = 'plot')
+
+spatFeatPlot2D(Xen_obj,
+ point_size = 3.5,
+ expression_values = 'plot',
+ show_image = F,
+ feats = 'ERBB2')
Subset the registered Visium and plot same gene
-#get the extent of giotto points, xmin, xmax, ymin, ymax
-subset_extent <- ext(gpoints@spatVector)
-sub_visium <- subsetGiottoLocs(V_final,
- x_min = subset_extent[1],
- x_max = subset_extent[2],
- y_min = subset_extent[3],
- y_max = subset_extent[4])
-spatFeatPlot2D(sub_visium,
- point_size = 2,
- expression_values = 'scaled',
- show_image = F,
- feats = 'ERBB2')
+#get the extent of giotto points, xmin, xmax, ymin, ymax
+subset_extent <- ext(gpoints@spatVector)
+sub_visium <- subsetGiottoLocs(V_final,
+ x_min = subset_extent[1],
+ x_max = subset_extent[2],
+ y_min = subset_extent[3],
+ y_max = subset_extent[4])
+spatFeatPlot2D(sub_visium,
+ point_size = 2,
+ expression_values = 'scaled',
+ show_image = F,
+ feats = 'ERBB2')
13.6.4 Register post-Xenium H&E and IF image
For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics.
-HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
- imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
- micron = 0.2125)
-plot(HE_xen)
+HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
+ imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
+ micron = 0.2125)
+plot(HE_xen)
The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use.
-HE_xen <- HE_xen |> flip(direction = "vertical")
-gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
-plot(gimg_rast)
-#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
+HE_xen <- HE_xen |> flip(direction = "vertical")
+gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
+plot(gimg_rast)
+#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots
-gg <- ggplot2::ggplot()
-pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
-pl + ggplot2::geom_smooth() +
- ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+gg <- ggplot2::ggplot()
+pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
+pl + ggplot2::geom_smooth() +
+ ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.4.1 Add registered image information and compare RNA vs protein expression
With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF
-CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
-HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
-Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
- largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
+CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
+HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
+Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
+ largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
Get the cell polygons, as Xenium and IF are both subcellular resolution
-cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
-colnames(cellpoly_dt) <- c('poly_ID','x','y')
-cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
-Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
+cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
+colnames(cellpoly_dt) <- c('poly_ID','x','y')
+cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
+Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
Compute the gene expression matrix by overlay the cell polygons and giotto points.
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'cell')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "cell",
- feat_info = "rna",
- aggr_function = "sum")
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'rna',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'ERBB2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'cell')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "cell",
+ feat_info = "rna",
+ aggr_function = "sum")
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'rna',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'ERBB2',
+ point_size = 0.05)
Now we overlay the HER2 expression from the raster image with the cell polygons.
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'cell',
- image_names = c('HER2','CD20'))
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'HER2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'cell',
+ image_names = c('HER2','CD20'))
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'HER2',
+ point_size = 0.05)
We can also overlay the protein expression to Visium spots
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'visium',
- image_names = c('HER2','CD20'))
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "visium",
- feat_info = "protein",
- aggr_function = "sum")
-
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'visium',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'visium',
- feats = 'HER2',
- point_size = 2)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'visium',
+ image_names = c('HER2','CD20'))
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "visium",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'visium',
+ feats = 'HER2',
+ point_size = 2)
@@ -944,29 +944,29 @@ 13.6.4.1 Add registered image inf
13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation
Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially.
Here, we provide an example of two compressed images to show the automatic alignment pipeline.
-
+
-IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
-terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
+IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
+terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform.
-estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
- flip_horizontal = T,
- use_single_channel = T,
- single_channel_number = 3),
- y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
- invert = T),
- plot_match = T,
- max_ratio = 0.5,estimate_fun = 'Projective')
+estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
+ flip_horizontal = T,
+ use_single_channel = T,
+ single_channel_number = 3),
+ y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
+ invert = T),
+ plot_match = T,
+ max_ratio = 0.5,estimate_fun = 'Projective')
Use the estimation, we can quickly visualize the transformation
-mtx <- as.matrix(estimation$params)
-transformed <- affine(IF, mtx)
-To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
-plot(HE)
-plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
-SIFT_registration_overlay
+mtx <- as.matrix(estimation$params)
+transformed <- affine(IF, mtx)
+To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
+plot(HE)
+plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
+SIFT_registration_overlay
diff --git a/spatial-proteomics-multiplexed-immunofluorescence.html b/spatial-proteomics-multiplexed-immunofluorescence.html
index 684d4e8..f550962 100644
--- a/spatial-proteomics-multiplexed-immunofluorescence.html
+++ b/spatial-proteomics-multiplexed-immunofluorescence.html
@@ -518,33 +518,33 @@ 11 Spatial proteomics: Multiplexe
Junxiang Xu
August 6th 2024
Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so.
-#Install the Giotto Environment with Cellpose, note that we only need to do it once
-reticulate::conda_create(envname = "giotto_cellpose",
- python_version = 3.8)
-#.re.restartR()
-reticulate::use_condaenv('giotto_cellpose')
-reticulate::py_install(
- pip = TRUE,
- envname = 'giotto_cellpose',
- packages = c(
- "pandas",
- "networkx",
- "python-igraph",
- "leidenalg",
- "scikit-learn",
- "cellpose",
- "smfishhmrf",
- 'tifffile',
- 'scikit-image'
- )
-)
-#.rs.restartR()
+#Install the Giotto Environment with Cellpose, note that we only need to do it once
+reticulate::conda_create(envname = "giotto_cellpose",
+ python_version = 3.8)
+#.re.restartR()
+reticulate::use_condaenv('giotto_cellpose')
+reticulate::py_install(
+ pip = TRUE,
+ envname = 'giotto_cellpose',
+ packages = c(
+ "pandas",
+ "networkx",
+ "python-igraph",
+ "leidenalg",
+ "scikit-learn",
+ "cellpose",
+ "smfishhmrf",
+ 'tifffile',
+ 'scikit-image'
+ )
+)
+#.rs.restartR()
Now, activate the Giotto python environment.
-#.rs.restartR()
-# Activate the Giotto python environment of your choice
-GiottoClass::set_giotto_python_path('giotto_cellpose')
-# Check if cellpose was successfully installed
-GiottoUtils::package_check('cellpose',repository = 'pip')
+#.rs.restartR()
+# Activate the Giotto python environment of your choice
+GiottoClass::set_giotto_python_path('giotto_cellpose')
+# Check if cellpose was successfully installed
+GiottoUtils::package_check('cellpose',repository = 'pip')
11.1 Spatial Proteomics Technologies
This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix.
@@ -557,58 +557,58 @@
11.2 Raw data type coming out of
11.2.1 Use ome.tiff as an example output data to begin with
OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore.
-library(Giotto)
-instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
- save_plot = TRUE,
- show_plot = TRUE,
- python_path = 'giotto_cellpose')
-options(timeout = 9999999)
-download_dir =file.path(getwd(),'/data/02_session4/')
-destfile = file.path(download_dir,'Lunaphore.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
-list.files(paste0(download_dir,'/Lunaphore'))
+library(Giotto)
+instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ python_path = 'giotto_cellpose')
+options(timeout = 9999999)
+download_dir =file.path(getwd(),'/data/02_session4/')
+destfile = file.path(download_dir,'Lunaphore.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
+list.files(paste0(download_dir,'/Lunaphore'))
We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML.
-img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
-img_meta
+img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
+img_meta
However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it.
-##Get channel information
-Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
-node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
-channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
-channel_df
+##Get channel information
+Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
+node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
+channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
+channel_df
11.2.2 Use single channel images as an example output data to begin with
Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like.
-#Create multichannel raster and extract each single channels
-Luna_terra = terra::rast(Luna)
-names(Luna_terra) <- channel_df$Name
-gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
-plot(gimg_DAPI)
+#Create multichannel raster and extract each single channels
+Luna_terra = terra::rast(Luna)
+names(Luna_terra) <- channel_df$Name
+gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
+plot(gimg_DAPI)
Extract and save the raster image for future use.
-single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
-if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
-for (i in 1:nrow(channel_df)){
- single_channel <- terra::subset(Luna_terra, i)
- terra::writeRaster(single_channel,
- filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
- overwrite=T)
-}
+single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
+if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
+for (i in 1:nrow(channel_df)){
+ single_channel <- terra::subset(Luna_terra, i)
+ terra::writeRaster(single_channel,
+ filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
+ overwrite=T)
+}
Create a list of GiottoLargeImages using single channel rasters.
-file_names = list.files(single_channel_dir,full.names = TRUE)
-image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
-gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
- names = image_names,
- negative_y = F,
- flip_vertical = T)
-names(gimg_list) <- image_names
-plot(gimg_list[['Vimentin']])
+file_names = list.files(single_channel_dir,full.names = TRUE)
+image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
+gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
+ names = image_names,
+ negative_y = F,
+ flip_vertical = T)
+names(gimg_list) <- image_names
+plot(gimg_list[['Vimentin']])
@@ -618,34 +618,34 @@ 11.3 Cell Segmentation file to ge
11.3.1 Using segmentation output file from DeepCell(mesmer) as an example.
We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function.
-gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
-plot(gpoly_mesmer)
+gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
+plot(gpoly_mesmer)
We can also zoom in to check how does the segmentation look.
-zoom <- c(2000,2500,2000,2500)
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+zoom <- c(2000,2500,2000,2500)
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
11.3.2 Using Giotto wrapper of Cellpose to perform segmentation
-gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
-writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
-#Create a giotto image to evaluate segmentation
-gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
+gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
+writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
+#Create a giotto image to evaluate segmentation
+gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website
-doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
- mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- channel_1 = 0,
- channel_2 = 0,
- model_name = 'cyto3',
- batch_size=12)
-cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- shift_horizontal_step = F,
- shift_vertical_step = F,
- flip_vertical = T)
-plot(gimg_for_cellpose)
-plot(cpoly, add = T, border = 'red')
+doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
+ mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ channel_1 = 0,
+ channel_2 = 0,
+ model_name = 'cyto3',
+ batch_size=12)
+cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ shift_horizontal_step = F,
+ shift_vertical_step = F,
+ flip_vertical = T)
+plot(gimg_for_cellpose)
+plot(cpoly, add = T, border = 'red')
@@ -654,133 +654,133 @@ 11.4 Create a Giotto Object using
You will need to have
- list of giotto images
- giottoPolygon created from segmentation
-Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
- images = gimg_list,
- instructions = instrs)
-Lunaphore_giotto
+Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
+ images = gimg_list,
+ instructions = instrs)
+Lunaphore_giotto
11.4.1 Overlap to matrix
calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with
-Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
- spatial_info = 'cell',
- image_names = names(gimg_list))
-
-Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
- type ='intensity',
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-showGiottoExpression(Lunaphore_giotto)
+Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
+ spatial_info = 'cell',
+ image_names = names(gimg_list))
+
+Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
+ type ='intensity',
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+showGiottoExpression(Lunaphore_giotto)
11.4.2 Manipulate Expression information
For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile.
Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object
-expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
-filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
-Lunaphore_giotto <- setExpression(Lunaphore_giotto,
- feat_type = 'protein',
- x = createExprObj(filtered_expr_mtx),
- name = 'raw')
-showGiottoExpression(Lunaphore_giotto)
+expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
+filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
+Lunaphore_giotto <- setExpression(Lunaphore_giotto,
+ feat_type = 'protein',
+ x = createExprObj(filtered_expr_mtx),
+ name = 'raw')
+showGiottoExpression(Lunaphore_giotto)
11.4.3 Rescale polygons
rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis
-Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
- poly_info = 'cell',
- name = 'smallcell',
- fx = 0.7, fy = 0.7, calculate_centroids = T)
-smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
-plot(smallpoly,add = TRUE, border = "red",ext = zoom)
+Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
+ poly_info = 'cell',
+ name = 'smallcell',
+ fx = 0.7, fy = 0.7, calculate_centroids = T)
+smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+plot(smallpoly,add = TRUE, border = "red",ext = zoom)
11.4.4 Perform clustering and differential expression
The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering
-Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
-Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
-
-Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
- scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
- set_seed = TRUE)
-screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
+Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
+Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
+
+Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
+ scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
+ set_seed = TRUE)
+screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check.
-Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
- dimensions_to_use = 1:5,
- set_seed = TRUE)
-Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
-Lunaphore_giotto = doHclust(Lunaphore_giotto,
- k = 8,
- dim_reduction_to_use = 'cells',
- spat_unit = 'cell',
- feat_type = 'protein')
-
-spatInSituPlotPoints(gobject = Lunaphore_giotto,
- spat_unit = "cell",
- polygon_feat_type = "cell",
- show_polygon = T,
- feat_type = "protein",
- feats = NULL,
- polygon_fill = 'hclust',
- polygon_fill_as_factor = TRUE,
- polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
- polygon_color = 'black',
- background_color = "white")
+Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
+ dimensions_to_use = 1:5,
+ set_seed = TRUE)
+Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
+Lunaphore_giotto = doHclust(Lunaphore_giotto,
+ k = 8,
+ dim_reduction_to_use = 'cells',
+ spat_unit = 'cell',
+ feat_type = 'protein')
+
+spatInSituPlotPoints(gobject = Lunaphore_giotto,
+ spat_unit = "cell",
+ polygon_feat_type = "cell",
+ show_polygon = T,
+ feat_type = "protein",
+ feats = NULL,
+ polygon_fill = 'hclust',
+ polygon_fill_as_factor = TRUE,
+ polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
+ polygon_color = 'black',
+ background_color = "white")
Then we can check the heatmap of protein expression and determine the first round of cluster annotation
-cluster_column = 'hclust'
-plotMetaDataHeatmap(Lunaphore_giotto,
- spat_unit = 'cell',feat_type = 'protein',
- expression_values = "raw",
- metadata_cols = c(cluster_column),
- selected_feats = names(gimg_list),
- y_text_size = 8,
- show_values = 'zscores_rescaled')
+cluster_column = 'hclust'
+plotMetaDataHeatmap(Lunaphore_giotto,
+ spat_unit = 'cell',feat_type = 'protein',
+ expression_values = "raw",
+ metadata_cols = c(cluster_column),
+ selected_feats = names(gimg_list),
+ y_text_size = 8,
+ show_values = 'zscores_rescaled')
11.4.5 Give the cluster an annotation based on expression values
-
+
11.4.6 spatial network
This is to create a cellular neighborhood based on nearest neighbor of physical distance
-Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
-
-spatPlot2D(Lunaphore_giotto,
- show_network = T,
- network_color = 'blue',
- point_size = 1.5,
- cell_color = 'hclust')
+Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
+
+spatPlot2D(Lunaphore_giotto,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1.5,
+ cell_color = 'hclust')
11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions
This is using cellProximityEnrichment() to statistically identify cell type interactions
-cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
- cluster_column = 'cell_types',
- spatial_network_name = 'Delaunay_network',
- adjust_method = 'fdr',
- number_of_simulations = 2000)
-## barplot
-cellProximityBarplot(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- min_orig_ints = 5, min_sim_ints = 5)
+cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
+ cluster_column = 'cell_types',
+ spatial_network_name = 'Delaunay_network',
+ adjust_method = 'fdr',
+ number_of_simulations = 2000)
+## barplot
+cellProximityBarplot(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ min_orig_ints = 5, min_sim_ints = 5)
-## network
-cellProximityNetwork(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- remove_self_edges = T,
- only_show_enrichment_edges = F)
+## network
+cellProximityNetwork(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ remove_self_edges = T,
+ only_show_enrichment_edges = F)
diff --git a/visium-hd.html b/visium-hd.html
index 23aa210..f24cdd4 100644
--- a/visium-hd.html
+++ b/visium-hd.html
@@ -525,7 +525,7 @@ 9.1 Objective9.2 Background
9.2.1 Visium HD Technology
-
+
Figure 9.1: Overview of Visium HD. Source: 10X Genomics
@@ -536,7 +536,7 @@
9.2.1 Visium HD Technology
9.2.2 Colorectal Cancer Sample
-
+
Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics
@@ -550,7 +550,7 @@
9.2.2 Colorectal Cancer Sample9.3 Data Ingestion
9.3.1 Visium HD output data format
-
+
Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline.
@@ -560,19 +560,19 @@
9.3.1 Visium HD output data forma
9.4 Tiling and aggregation
The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells.
-
+
9.5 Scalability and projection functions
diff --git a/visium-part-i.html b/visium-part-i.html
index 5aa1658..c4746ee 100644
--- a/visium-part-i.html
+++ b/visium-part-i.html
@@ -520,7 +520,7 @@ 7 Visium Part I
7.1 The Visium technology
Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression.
-
+
Figure 7.1: Visum workflow. Source: 10X Genomics
@@ -536,73 +536,73 @@
7.2 Introduction to the spatial d
7.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/01_session5/")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
+dir.create("data/01_session5/")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”.
-
+
7.4 Create the Giotto object
createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/01_session5/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-## Provide the path to the visium folder
-data_path <- "data/01_session5/"
-
-## Create object directly from the visium folder
-visium_brain <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/01_session5/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+## Provide the path to the visium folder
+data_path <- "data/01_session5/"
+
+## Create object directly from the visium folder
+visium_brain <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
7.5 Subset on spots that were covered by tissue
Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area.
-spatPlot2D(
- gobject = visium_brain,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE)
-
+spatPlot2D(
+ gobject = visium_brain,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE)
+
Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area.
-
+
7.6 Quality control
@@ -610,43 +610,43 @@ 7.6 Quality controlvisium_brain_statistics <- addStatistics(gobject = visium_brain,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = visium_brain_statistics,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain_statistics <- addStatistics(gobject = visium_brain,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = visium_brain_statistics,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.3: Spatial distribution of features per spot.
filterDistributions() creates a histogram to show the distribution of features per spot across the sample.
-
-
+
+
Figure 7.4: Distribution of features per spot.
When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample.
-
-
+
+
Figure 7.5: Distribution of cells with different features per spot.
filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data:
-filterCombinations(gobject = visium_brain_statistics,
- expression_thresholds = c(1, 2, 3),
- feat_det_in_min_cells = c(50, 100, 200),
- min_det_feats_per_cell = c(500, 1000, 1500))
-
+filterCombinations(gobject = visium_brain_statistics,
+ expression_thresholds = c(1, 2, 3),
+ feat_det_in_min_cells = c(50, 100, 200),
+ min_det_feats_per_cell = c(500, 1000, 1500))
+
Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations.
@@ -656,34 +656,34 @@
7.6 Quality control
7.7 Filtering
Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample.
-visium_brain <- filterGiotto(
- gobject = visium_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE
-)
-Feature type: rna
-Number of cells removed: 4 out of 2702
-Number of feats removed: 7311 out of 22125
+
+
7.8 Normalization
Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one.
-visium_brain <- normalizeGiotto(
- gobject = visium_brain,
- scalefactor = 6000,
- verbose = TRUE
-)
+visium_brain <- normalizeGiotto(
+ gobject = visium_brain,
+ scalefactor = 6000,
+ verbose = TRUE
+)
Calculate the normalized number of features per spot and save the statistics in the metadata table.
-visium_brain <- addStatistics(gobject = visium_brain)
-
-## visualize
-spatPlot2D(gobject = visium_brain,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain <- addStatistics(gobject = visium_brain)
+
+## visualize
+spatPlot2D(gobject = visium_brain,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.7: Spatial distribution of the number of features per spot.
@@ -698,11 +698,11 @@
7.9.1 Highly Variable Features:
loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_loess",
- save_plot = TRUE,
- default_save_name = "HVFplot_loess")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_loess",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_loess")
+
Figure 7.8: Covariance of HVFs using the loess method.
@@ -711,11 +711,11 @@
7.9.1 Highly Variable Features:
pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")
+
Figure 7.9: Variance of HVFs using the pearson residuals method.
@@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
17.4.1 Read in the poly information
First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object.
-# Read in the data
-stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson")
-
-stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path,
- name = "stardist_cell",
- calc_centroids = TRUE)
-
-stardist_cell_gpoly <- flip(stardist_cell_gpoly)# Read in the data
+stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson")
+
+stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path,
+ name = "stardist_cell",
+ calc_centroids = TRUE)
+
+stardist_cell_gpoly <- flip(stardist_cell_gpoly)17.4.2 Vizualizing polygons
Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain.
- -
+
+
-
Figure 17.3: Mouse brain cell polygons from the visium dataset
+
17.4.2 Vizualizing polygons
Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped @@ -686,8 +686,8 @@
17.4.2 Vizualizing polygons
17.4.3 Showing Giotto object prior to polygon addition
Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon.
-
-
+
+
Figure 17.5: Giotto object before adding subcellular polygons.
@@ -697,11 +697,11 @@
17.4.3 Showing Giotto object prio
17.4.4 Adding polygons to giotto object
After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject.
-v_brain <- addGiottoPolygons(v_brain,
- gpolygons = list('stardist_cell' = stardist_cell_gpoly))
-
-print(v_brain)
-
+v_brain <- addGiottoPolygons(v_brain,
+ gpolygons = list('stardist_cell' = stardist_cell_gpoly))
+
+print(v_brain)
+
Figure 17.6: Giotto object after to adding subcellular polygons.
@@ -711,9 +711,9 @@
17.4.4 Adding polygons to giotto
17.4.5 Check polygon information
We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots.
-
-
+
+
Figure 17.7: Polygon information for stardist_cell.
@@ -727,23 +727,23 @@
17.5 Performing kriging17.5.1 Interpolating features
Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be
analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required.
-future::plan(future::multisession()) # comment out for single threading
-subcell_v_brain <- interpolateFeature(v_brain,
- spat_unit = 'cell',
- feat_type = 'rna',
- ext = ext(v_brain),
- feats = ext_spatial_features,
- overwrite = T)
-
-print(subcell_v_brain)
-
+future::plan(future::multisession()) # comment out for single threading
+subcell_v_brain <- interpolateFeature(v_brain,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ ext = ext(v_brain),
+ feats = ext_spatial_features,
+ overwrite = T)
+
+print(subcell_v_brain)
+
Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right).
For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1.
-
+
Figure 17.9: Raster of gene expression interpolation for Pantr1
@@ -754,18 +754,18 @@
17.5.1 Interpolating features17.5.2 Expression overlap
The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier.
This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run.
-subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain,
- name_overlap = "rna",
- spatial_info = "stardist_cell",
- image_names = ext_spatial_features)
-
-subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain,
- poly_info = "stardist_cell",
- feat_info = "rna",
- aggr_function = "sum",
- type='intensity')
+subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain,
+ name_overlap = "rna",
+ spatial_info = "stardist_cell",
+ image_names = ext_spatial_features)
+
+subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain,
+ poly_info = "stardist_cell",
+ feat_info = "rna",
+ aggr_function = "sum",
+ type='intensity')
After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified.
-
+
Figure 17.10: Gene expression for cells based on interpolation.
@@ -776,70 +776,70 @@
17.5.2 Expression overlap
17.6 Reading in larger dataset
For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here.
-
+
17.7 Analyzing interpolated features
17.7.1 Filter and normalization
Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis.
-subcell_v_brain <- filterGiotto(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- expression_values = "raw",
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-
-subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- scalefactor = 6000,
- verbose = T)
+subcell_v_brain <- filterGiotto(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ expression_values = "raw",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+
+subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ scalefactor = 6000,
+ verbose = T)
17.7.2 Visualizing gene expression from interpolated expression
Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data.
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Sparc',
- point_size = 2,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_sparc_vis'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'stardist_cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Sparc',
- point_size = 0.6,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_sparc'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'cell',
- gradient_style = 'sequential',
- feats = 'Pantr1',
- cell_color_gradient = 'Geyser',
- point_size = 2,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_pantr1_vis'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'stardist_cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Pantr1',
- point_size = 0.6,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_pantr1'))
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Sparc',
+ point_size = 2,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_sparc_vis'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Sparc',
+ point_size = 0.6,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_sparc'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'cell',
+ gradient_style = 'sequential',
+ feats = 'Pantr1',
+ cell_color_gradient = 'Geyser',
+ point_size = 2,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_pantr1_vis'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Pantr1',
+ point_size = 0.6,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_pantr1'))
Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer.
-
+
Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom).
@@ -848,37 +848,37 @@
17.7.2 Visualizing gene expressio
17.7.3 Run PCA
-
+
17.7.4 Clustering
-# UMAP
-subcell_v_brain <- runUMAP(subcell_v_brain,
- spat_unit = "stardist_cell",
- dimensions_to_use = 1:15,
- n_neighbors = 1000,
- min_dist = 0.001,
- spread = 1)
-
-# NN Network
-subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- dimensions_to_use = 1:10,
- feats_to_use = hv_feats,
- expression_values = 'normalized',
- k = 70)
-
-subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- resolution = 0.15,
- n_iterations = 100,
- partition_type = 'RBConfigurationVertexPartition')
-
-plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus')
-
+# UMAP
+subcell_v_brain <- runUMAP(subcell_v_brain,
+ spat_unit = "stardist_cell",
+ dimensions_to_use = 1:15,
+ n_neighbors = 1000,
+ min_dist = 0.001,
+ spread = 1)
+
+# NN Network
+subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ dimensions_to_use = 1:10,
+ feats_to_use = hv_feats,
+ expression_values = 'normalized',
+ k = 70)
+
+subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ resolution = 0.15,
+ n_iterations = 100,
+ partition_type = 'RBConfigurationVertexPartition')
+
+plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus')
+
Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering.
@@ -888,27 +888,27 @@
17.7.4 Clustering
17.7.5 Visualizing clustering
Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster.
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.5,
- point_shape = 'no_border',
- background_color = 'black',
- save_plot = F,
- show_legend = TRUE)
-
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat'))
-
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.5,
+ point_shape = 'no_border',
+ background_color = 'black',
+ save_plot = F,
+ show_legend = TRUE)
+
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat'))
+
Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
@@ -918,37 +918,37 @@
17.7.5 Visualizing clustering
17.7.6 Cropping objects
We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below.
-subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
- spat_unit = ":all:",
- x_min = 4000,
- x_max = 7000,
- y_min = -6500,
- y_max = -3500,
- z_max = NULL,
- z_min = NULL)
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 2,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_vis_spat_crop'))
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat_crop'))
-
+subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
+ spat_unit = ":all:",
+ x_min = 4000,
+ x_max = 7000,
+ y_min = -6500,
+ y_max = -3500,
+ z_max = NULL,
+ z_min = NULL)
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 2,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_vis_spat_crop'))
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat_crop'))
+
Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
diff --git a/interoperability-with-isolated-tools.html b/interoperability-with-isolated-tools.html
index ceb2d0a..8cca7e6 100644
--- a/interoperability-with-isolated-tools.html
+++ b/interoperability-with-isolated-tools.html
@@ -526,1521 +526,1521 @@
16.1.1 Dataset downloaddata_path <- "data"
-dir.create(data_path)
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
- destfile = file.path(data_path,"counts.h5ad"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
- destfile = file.path(data_path,"cell_labels.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
+data_path <- "data"
+dir.create(data_path)
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
+ destfile = file.path(data_path,"counts.h5ad"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
+ destfile = file.path(data_path,"cell_labels.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
16.1.2 Create the Giotto object
-library(Giotto)
-library(reticulate)
-
-## Set instructions
-results_folder <- "results"
-dir.create(results_folder)
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-
-## load data
-
-### meta data
-meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-colnames(meta_df)[[1]] <- "cell_ID"
-slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
-slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
-
-### counts matrix
-scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
-ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
-reticulate::source_python(ad2g_path)
-adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
-X <- extract_expression(adata)
-cID <- extract_cell_IDs(adata)
-fID <- extract_feat_IDs(adata)
-rownames(X) <- fID
-colnames(X) <- cID
-slice1_filter <- cID %in% slice1_meta_df$cell_ID
-slice2_filter <- cID %in% slice2_meta_df$cell_ID
-slice1_exp <- X[,slice1_filter]
-slice2_exp <- X[,slice2_filter]
-
-### cell segmentation to cell location
-segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_df <- rbind(segments_1_df, segments_2_df)
-loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
-loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
-loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
-centr.x <- apply(loc.use[,loc.x],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-centr.y <- apply(loc.use[,loc.y],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
-slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
-slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
-
-## giotto obj
-giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
- spatial_locs = slice1_spatial_locs_df,
- cell_metadata = slice1_meta_df)
-giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
- spatial_locs = slice2_spatial_locs_df,
- cell_metadata = slice2_meta_df)
-giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
- gobject_names = c('mouse2_slice229',
- 'mouse2_slice300'), # name for each samples
- join_method = 'z_stack')
-# We skip the processing process here to save time and use the given cell type
-# annotation directly
-
-ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
- cell_type = "subclass",
- output_path = data_path,
- spat_unit = "cell",
- feat_type = "rna",
- verbose = TRUE)
-
-# Cell_ID Sample x y Cell_Type
-# <char> <char> <dbl> <dbl> <char>
-# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
-# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
-# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
-# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
-# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
-# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
+library(Giotto)
+library(reticulate)
+
+## Set instructions
+results_folder <- "results"
+dir.create(results_folder)
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+
+## load data
+
+### meta data
+meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+colnames(meta_df)[[1]] <- "cell_ID"
+slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
+slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
+
+### counts matrix
+scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
+ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
+reticulate::source_python(ad2g_path)
+adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
+X <- extract_expression(adata)
+cID <- extract_cell_IDs(adata)
+fID <- extract_feat_IDs(adata)
+rownames(X) <- fID
+colnames(X) <- cID
+slice1_filter <- cID %in% slice1_meta_df$cell_ID
+slice2_filter <- cID %in% slice2_meta_df$cell_ID
+slice1_exp <- X[,slice1_filter]
+slice2_exp <- X[,slice2_filter]
+
+### cell segmentation to cell location
+segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_df <- rbind(segments_1_df, segments_2_df)
+loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
+loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
+loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
+centr.x <- apply(loc.use[,loc.x],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+centr.y <- apply(loc.use[,loc.y],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
+slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
+slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
+
+## giotto obj
+giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
+ spatial_locs = slice1_spatial_locs_df,
+ cell_metadata = slice1_meta_df)
+giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
+ spatial_locs = slice2_spatial_locs_df,
+ cell_metadata = slice2_meta_df)
+giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
+ gobject_names = c('mouse2_slice229',
+ 'mouse2_slice300'), # name for each samples
+ join_method = 'z_stack')
+# We skip the processing process here to save time and use the given cell type
+# annotation directly
+
+ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
+ cell_type = "subclass",
+ output_path = data_path,
+ spat_unit = "cell",
+ feat_type = "rna",
+ verbose = TRUE)
+
+# Cell_ID Sample x y Cell_Type
+# <char> <char> <dbl> <dbl> <char>
+# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
+# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
+# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
+# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
+# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
+# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
Spatial cell type distributions
-spatPlot2D(giotto_obj,
- group_by = "slice_id",
- cell_color = "subclass",
- point_size = 1,
- point_border_stroke = NA,
- legend_text = 6)
+spatPlot2D(giotto_obj,
+ group_by = "slice_id",
+ cell_color = "subclass",
+ point_size = 1,
+ point_border_stroke = NA,
+ legend_text = 6)
16.1.2.1 Installation ONTraC
You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node.
It will run for less than 10 minutes on this example dataset.
For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time.
-source ~/.bash_profile
-conda create -y -n ONTraC python=3.11
-conda activate ONTraC
-pip install git+https://github.com/gyuanlab/ONTraC.git@V2
-# Retrieving notices: ...working... done
-# Channels:
-# - conda-forge
-# - bioconda
-# - r
-# - pytorch
-# - defaults
-# Platform: osx-arm64
-# Collecting package metadata (repodata.json): ...working... done
-# Solving environment: ...working... done
-#
-# ## Package Plan ##
-#
-# environment location: /Users/wwang/miniconda3/envs/ONTraC
-#
-# added / updated specs:
-# - python=3.11
-#
-#
-# The following packages will be downloaded:
-#
-# package | build
-# ---------------------------|-----------------
-# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
-# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
-# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
-# ------------------------------------------------------------
-# Total: 4.3 MB
-#
-# The following NEW packages will be INSTALLED:
-#
-# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
-# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
-# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
-# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
-# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
-# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
-# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
-# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
-# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
-# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
-# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
-# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
-# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
-# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
-# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
-# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
-#
-#
-#
-# Downloading and Extracting Packages: ...working... done
-# Preparing transaction: ...working... done
-# Verifying transaction: ...working... done
-# Executing transaction: ...working... done
-# #
-# # To activate this environment, use
-# #
-# # $ conda activate ONTraC
-# #
-# # To deactivate an active environment, use
-# #
-# # $ conda deactivate
-#
-# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
-# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
-# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
-# Running command git checkout -b V2 --track origin/V2
-# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
-# Installing build dependencies: started
-# Switched to a new branch 'V2'
-# branch 'V2' set up to track 'origin/V2'.
-# Installing build dependencies: finished with status 'done'
-# Getting requirements to build wheel: started
-# Getting requirements to build wheel: finished with status 'done'
-# Preparing metadata (pyproject.toml): started
-# Preparing metadata (pyproject.toml): finished with status 'done'
-# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
-# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
-# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
-# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
-# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
-# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
-# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
-# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
-# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
-# Collecting session-info (from ONTraC==2.0a9.dev7)
-# Using cached session_info-1.0.0-py3-none-any.whl
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
-# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
-# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
-# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
-# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
-# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
-# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
-# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
-# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
-# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
-# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
-# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
-# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
-# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
-# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
-# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
-# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
-# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
-# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
-# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
-# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
-# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
-# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
-# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
-# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
-# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
-# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
-# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
-# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
-# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
-# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
-# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
-# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
-# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
-# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
-# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
-# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
-# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
-# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
-# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
-# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
-# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
-# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
-# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
-# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
-# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
-# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
-# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
-# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
-# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
-# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
-# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
-# Using cached idna-3.7-py3-none-any.whl (66 kB)
-# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
-# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
-# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
-# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
-# Building wheels for collected packages: ONTraC
-# Building wheel for ONTraC (pyproject.toml): started
-# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
-# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
-# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
-# Successfully built ONTraC
-# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
-# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
+source ~/.bash_profile
+conda create -y -n ONTraC python=3.11
+conda activate ONTraC
+pip install git+https://github.com/gyuanlab/ONTraC.git@V2
+# Retrieving notices: ...working... done
+# Channels:
+# - conda-forge
+# - bioconda
+# - r
+# - pytorch
+# - defaults
+# Platform: osx-arm64
+# Collecting package metadata (repodata.json): ...working... done
+# Solving environment: ...working... done
+#
+# ## Package Plan ##
+#
+# environment location: /Users/wwang/miniconda3/envs/ONTraC
+#
+# added / updated specs:
+# - python=3.11
+#
+#
+# The following packages will be downloaded:
+#
+# package | build
+# ---------------------------|-----------------
+# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
+# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
+# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
+# ------------------------------------------------------------
+# Total: 4.3 MB
+#
+# The following NEW packages will be INSTALLED:
+#
+# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
+# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
+# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
+# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
+# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
+# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
+# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
+# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
+# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
+# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
+# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
+# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
+# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
+# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
+# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
+# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
+#
+#
+#
+# Downloading and Extracting Packages: ...working... done
+# Preparing transaction: ...working... done
+# Verifying transaction: ...working... done
+# Executing transaction: ...working... done
+# #
+# # To activate this environment, use
+# #
+# # $ conda activate ONTraC
+# #
+# # To deactivate an active environment, use
+# #
+# # $ conda deactivate
+#
+# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
+# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
+# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
+# Running command git checkout -b V2 --track origin/V2
+# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
+# Installing build dependencies: started
+# Switched to a new branch 'V2'
+# branch 'V2' set up to track 'origin/V2'.
+# Installing build dependencies: finished with status 'done'
+# Getting requirements to build wheel: started
+# Getting requirements to build wheel: finished with status 'done'
+# Preparing metadata (pyproject.toml): started
+# Preparing metadata (pyproject.toml): finished with status 'done'
+# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
+# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
+# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
+# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
+# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
+# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
+# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
+# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
+# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
+# Collecting session-info (from ONTraC==2.0a9.dev7)
+# Using cached session_info-1.0.0-py3-none-any.whl
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
+# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
+# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
+# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
+# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
+# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
+# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
+# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
+# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
+# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
+# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
+# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
+# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
+# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
+# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
+# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
+# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
+# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
+# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
+# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
+# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
+# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
+# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
+# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
+# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
+# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
+# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
+# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
+# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
+# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
+# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
+# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
+# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
+# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
+# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
+# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
+# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
+# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
+# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
+# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
+# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
+# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
+# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
+# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
+# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
+# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
+# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
+# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
+# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
+# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
+# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
+# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
+# Using cached idna-3.7-py3-none-any.whl (66 kB)
+# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
+# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
+# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
+# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
+# Building wheels for collected packages: ONTraC
+# Building wheel for ONTraC (pyproject.toml): started
+# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
+# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
+# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
+# Successfully built ONTraC
+# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
+# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
16.1.2.2 Running ONTraC
-
-source ~/.bash_profile
-conda activate ONTraC_v2_py311
-ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
-# ##################################################################################
-#
-# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
-# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
-# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
-# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
-# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
-#
-# version: 2.0a11
-#
-# ##################################################################################
-# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
-# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
-# 11:33:23 --- INFO: -------- I/O options -------
-# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
-# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
-# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
-# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
-# 11:33:23 --- INFO: -------- preprocessing options -------
-# 11:33:23 --- INFO: resolution: 10.0
-# 11:33:23 --- INFO: -------- niche net constr options -------
-# 11:33:23 --- INFO: n_cpu: 4
-# 11:33:23 --- INFO: n_neighbors: 50
-# 11:33:23 --- INFO: n_local: 20
-# 11:33:23 --- INFO: embedding_adjust: False
-# 11:33:23 --- INFO: sigma: 1
-# 11:33:23 --- INFO: -------- train options -------
-# 11:33:23 --- INFO: device: cpu
-# 11:33:23 --- INFO: epochs: 1000
-# 11:33:23 --- INFO: batch_size: 0
-# 11:33:23 --- INFO: patience: 100
-# 11:33:23 --- INFO: min_delta: 0.001
-# 11:33:23 --- INFO: min_epochs: 50
-# 11:33:23 --- INFO: seed: 42
-# 11:33:23 --- INFO: lr: 0.03
-# 11:33:23 --- INFO: hidden_feats: 4
-# 11:33:23 --- INFO: k: 6
-# 11:33:23 --- INFO: modularity_loss_weight: 1.0
-# 11:33:23 --- INFO: purity_loss_weight: 100.0
-# 11:33:23 --- INFO: regularization_loss_weight: 0.1
-# 11:33:23 --- INFO: beta: 0.3
-# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
-# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
-# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
-# 11:33:23 --- INFO: ------------- Niche network construct ---------------
-# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
-# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
-# 11:33:28 --- INFO: Generating samples.yaml file.
-# 11:33:28 --- INFO: ------------ Niche network construct end ------------
-# 11:33:28 --- INFO: ------------------------ GNN ------------------------
-# 11:33:28 --- INFO: Loading dataset.
-# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
-# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
-# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
-# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
-# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
-# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
-# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
-# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
-# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
-# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
-# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
-# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
-# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
-# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
-# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
-# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
-# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
-# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
-# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
-# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
-# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
-# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
-# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
-# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
-# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
-# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
-# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
-# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
-# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
-# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
-# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
-# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
-# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
-# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
-# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
-# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
-# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
-# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
-# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
-# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
-# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
-# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
-# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
-# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
-# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
-# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
-# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
-# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
-# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
-# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
-# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
-# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
-# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
-# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
-# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
-# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
-# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
-# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
-# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
-# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
-# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
-# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
-# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
-# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
-# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
-# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
-# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
-# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
-# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
-# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
-# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
-# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
-# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
-# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
-# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
-# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
-# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
-# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
-# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
-# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
-# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
-# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
-# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
-# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
-# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
-# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
-# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
-# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
-# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
-# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
-# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
-# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
-# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
-# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
-# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
-# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
-# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
-# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
-# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
-# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
-# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
-# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
-# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
-# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
-# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
-# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
-# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
-# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
-# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
-# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
-# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
-# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
-# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
-# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
-# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
-# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
-# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
-# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
-# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
-# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
-# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
-# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
-# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
-# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
-# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
-# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
-# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
-# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
-# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
-# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
-# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
-# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
-# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
-# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
-# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
-# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
-# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
-# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
-# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
-# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
-# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
-# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
-# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
-# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
-# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
-# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
-# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
-# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
-# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
-# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
-# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
-# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
-# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
-# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
-# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
-# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
-# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
-# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
-# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
-# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
-# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
-# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
-# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
-# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
-# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
-# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
-# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
-# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
-# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
-# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
-# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
-# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
-# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
-# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
-# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
-# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
-# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
-# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
-# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
-# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
-# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
-# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
-# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
-# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
-# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
-# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
-# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
-# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
-# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
-# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
-# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
-# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
-# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
-# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
-# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
-# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
-# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
-# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
-# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
-# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
-# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
-# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
-# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
-# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
-# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
-# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
-# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
-# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
-# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
-# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
-# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
-# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
-# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
-# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
-# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
-# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
-# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
-# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
-# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
-# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
-# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
-# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
-# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
-# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
-# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
-# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
-# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
-# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
-# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
-# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
-# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
-# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
-# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
-# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
-# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
-# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
-# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
-# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
-# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
-# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
-# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
-# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
-# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
-# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
-# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
-# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
-# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
-# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
-# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
-# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
-# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
-# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
-# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
-# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
-# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
-# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
-# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
-# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
-# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
-# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
-# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
-# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
-# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
-# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
-# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
-# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
-# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
-# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
-# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
-# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
-# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
-# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
-# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
-# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
-# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
-# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
-# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
-# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
-# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
-# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
-# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
-# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
-# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
-# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
-# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
-# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
-# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
-# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
-# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
-# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
-# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
-# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
-# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
-# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
-# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
-# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
-# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
-# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
-# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
-# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
-# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
-# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
-# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
-# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
-# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
-# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
-# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
-# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
-# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
-# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
-# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
-# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
-# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
-# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
-# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
-# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
-# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
-# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
-# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
-# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
-# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
-# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
-# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
-# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
-# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
-# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
-# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
-# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
-# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
-# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
-# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
-# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
-# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
-# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
-# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
-# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
-# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
-# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
-# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
-# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
-# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
-# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
-# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
-# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
-# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
-# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
-# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
-# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
-# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
-# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
-# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
-# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
-# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
-# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
-# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
-# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
-# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
-# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
-# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
-# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
-# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
-# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
-# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
-# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
-# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
-# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
-# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
-# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
-# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
-# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
-# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
-# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
-# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
-# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
-# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
-# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
-# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
-# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
-# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
-# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
-# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
-# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
-# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
-# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
-# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
-# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
-# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
-# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
-# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
-# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
-# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
-# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
-# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
-# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
-# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
-# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
-# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
-# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
-# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
-# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
-# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
-# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
-# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
-# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
-# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
-# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
-# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
-# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
-# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
-# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
-# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
-# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
-# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
-# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
-# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
-# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
-# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
-# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
-# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
-# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
-# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
-# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
-# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
-# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
-# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
-# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
-# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
-# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
-# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
-# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
-# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
-# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
-# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
-# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
-# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
-# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
-# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
-# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
-# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
-# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
-# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
-# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
-# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
-# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
-# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
-# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
-# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
-# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
-# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
-# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
-# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
-# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
-# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
-# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
-# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
-# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
-# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
-# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
-# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
-# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
-# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
-# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
-# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
-# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
-# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
-# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
-# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
-# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
-# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
-# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
-# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
-# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
-# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
-# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
-# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
-# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
-# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
-# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
-# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
-# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
-# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
-# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
-# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
-# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
-# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
-# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
-# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
-# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
-# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
-# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
-# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
-# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
-# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
-# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
-# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
-# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
-# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
-# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
-# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
-# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
-# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
-# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
-# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
-# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
-# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
-# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
-# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
-# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
-# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
-# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
-# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
-# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
-# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
-# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
-# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
-# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
-# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
-# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
-# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
-# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
-# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
-# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
-# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
-# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
-# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
-# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
-# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
-# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
-# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
-# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
-# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
-# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
-# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
-# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
-# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
-# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
-# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
-# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
-# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
-# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
-# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
-# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
-# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
-# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
-# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
-# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
-# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
-# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
-# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
-# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
-# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
-# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
-# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
-# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
-# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
-# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
-# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
-# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
-# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
-# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
-# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
-# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
-# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
-# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
-# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
-# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
-# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
-# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
-# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
-# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
-# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
-# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
-# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
-# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
-# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
-# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
-# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
-# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
-# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
-# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
-# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
-# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
-# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
-# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
-# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
-# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
-# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
-# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
-# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
-# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
-# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
-# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
-# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
-# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
-# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
-# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
-# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
-# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
-# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
-# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
-# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
-# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
-# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
-# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
-# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
-# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
-# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
-# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
-# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
-# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
-# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
-# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
-# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
-# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
-# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
-# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
-# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
-# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
-# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
-# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
-# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
-# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
-# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
-# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
-# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
-# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
-# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
-# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
-# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
-# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
-# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
-# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
-# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
-# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
-# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
-# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
-# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
-# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
-# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
-# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
-# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
-# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
-# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
-# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
-# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
-# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
-# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
-# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
-# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
-# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
-# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
-# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
-# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
-# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
-# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
-# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
-# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
-# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
-# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
-# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
-# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
-# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
-# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
-# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
-# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
-# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
-# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
-# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
-# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
-# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
-# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
-# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
-# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
-# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
-# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
-# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
-# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
-# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
-# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
-# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
-# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
-# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
-# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
-# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
-# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
-# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
-# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
-# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
-# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
-# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
-# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
-# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
-# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
-# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
-# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
-# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
-# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
-# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
-# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
-# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
-# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
-# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
-# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
-# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
-# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
-# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
-# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
-# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
-# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
-# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
-# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
-# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
-# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
-# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
-# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
-# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
-# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
-# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
-# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
-# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
-# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
-# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
-# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
-# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
-# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
-# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
-# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
-# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
-# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
-# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
-# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
-# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
-# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
-# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
-# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
-# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
-# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
-# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
-# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
-# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
-# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
-# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
-# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
-# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
-# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
-# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
-# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
-# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
-# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
-# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
-# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
-# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
-# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
-# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
-# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
-# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
-# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
-# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
-# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
-# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
-# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
-# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
-# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
-# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
-# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
-# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
-# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
-# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
-# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
-# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
-# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
-# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
-# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
-# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
-# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
-# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
-# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
-# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
-# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
-# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
-# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
-# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
-# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
-# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
-# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
-# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
-# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
-# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
-# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
-# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
-# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
-# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
-# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
-# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
-# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
-# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
-# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
-# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
-# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
-# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
-# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
-# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
-# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
-# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
-# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
-# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
-# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
-# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
-# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
-# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
-# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
-# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
-# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
-# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
-# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
-# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
-# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
-# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
-# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
-# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
-# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
-# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
-# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
-# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
-# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
-# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
-# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
-# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
-# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
-# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
-# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
-# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
-# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
-# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
-# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
-# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
-# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
-# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
-# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
-# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
-# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
-# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
-# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
-# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
-# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
-# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
-# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
-# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
-# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
-# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
-# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
-# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
-# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
-# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
-# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
-# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
-# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
-# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
-# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
-# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
-# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
-# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
-# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
-# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
-# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
-# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
-# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
-# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
-# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
-# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
-# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
-# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
-# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
-# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
-# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
-# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
-# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
-# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
-# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
-# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
-# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
-# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
-# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
-# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
-# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
-# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
-# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
-# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
-# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
-# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
-# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
-# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
-# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
-# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
-# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
-# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
-# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
-# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
-# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
-# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
-# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
-# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
-# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
-# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
-# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
-# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
-# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
-# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
-# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
-# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
-# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
-# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
-# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
-# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
-# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
-# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
-# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
-# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
-# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
-# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
-# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
-# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
-# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
-# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
-# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
-# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
-# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
-# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
-# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
-# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
-# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
-# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
-# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
-# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
-# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
-# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
-# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
-# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
-# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
-# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
-# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
-# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
-# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
-# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
-# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
-# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
-# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
-# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
-# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
-# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
-# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
-# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
-# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
-# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
-# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
-# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
-# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
-# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
-# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
-# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
-# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
-# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
-# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
-# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
-# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
-# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
-# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
-# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
-# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
-# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
-# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
-# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
-# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
-# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
-# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
-# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
-# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
-# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
-# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
-# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
-# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
-# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
-# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
-# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
-# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
-# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
-# 11:38:37 --- INFO: Training process end.
-# 11:38:37 --- INFO: Evaluating process start.
-# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
-# 11:38:37 --- INFO: Evaluating process end.
-# 11:38:37 --- INFO: Predicting process start.
-# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
-# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
-# 11:38:38 --- INFO: Predicting process end.
-# 11:38:38 --- INFO: --------------------- GNN end ----------------------
-# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
-# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
-# 11:38:38 --- INFO: Loading dataset.
-# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:38:39 --- INFO: Calculating NTScore for each niche.
-# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
-# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
-# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
-# 11:38:39 --- INFO: Output NTScore tables.
-# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
+
+source ~/.bash_profile
+conda activate ONTraC_v2_py311
+ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
+# ##################################################################################
+#
+# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
+# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
+# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
+# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
+# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
+#
+# version: 2.0a11
+#
+# ##################################################################################
+# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
+# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
+# 11:33:23 --- INFO: -------- I/O options -------
+# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
+# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
+# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
+# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
+# 11:33:23 --- INFO: -------- preprocessing options -------
+# 11:33:23 --- INFO: resolution: 10.0
+# 11:33:23 --- INFO: -------- niche net constr options -------
+# 11:33:23 --- INFO: n_cpu: 4
+# 11:33:23 --- INFO: n_neighbors: 50
+# 11:33:23 --- INFO: n_local: 20
+# 11:33:23 --- INFO: embedding_adjust: False
+# 11:33:23 --- INFO: sigma: 1
+# 11:33:23 --- INFO: -------- train options -------
+# 11:33:23 --- INFO: device: cpu
+# 11:33:23 --- INFO: epochs: 1000
+# 11:33:23 --- INFO: batch_size: 0
+# 11:33:23 --- INFO: patience: 100
+# 11:33:23 --- INFO: min_delta: 0.001
+# 11:33:23 --- INFO: min_epochs: 50
+# 11:33:23 --- INFO: seed: 42
+# 11:33:23 --- INFO: lr: 0.03
+# 11:33:23 --- INFO: hidden_feats: 4
+# 11:33:23 --- INFO: k: 6
+# 11:33:23 --- INFO: modularity_loss_weight: 1.0
+# 11:33:23 --- INFO: purity_loss_weight: 100.0
+# 11:33:23 --- INFO: regularization_loss_weight: 0.1
+# 11:33:23 --- INFO: beta: 0.3
+# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
+# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
+# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
+# 11:33:23 --- INFO: ------------- Niche network construct ---------------
+# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
+# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
+# 11:33:28 --- INFO: Generating samples.yaml file.
+# 11:33:28 --- INFO: ------------ Niche network construct end ------------
+# 11:33:28 --- INFO: ------------------------ GNN ------------------------
+# 11:33:28 --- INFO: Loading dataset.
+# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
+# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
+# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
+# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
+# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
+# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
+# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
+# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
+# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
+# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
+# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
+# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
+# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
+# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
+# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
+# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
+# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
+# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
+# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
+# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
+# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
+# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
+# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
+# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
+# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
+# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
+# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
+# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
+# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
+# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
+# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
+# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
+# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
+# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
+# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
+# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
+# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
+# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
+# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
+# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
+# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
+# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
+# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
+# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
+# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
+# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
+# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
+# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
+# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
+# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
+# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
+# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
+# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
+# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
+# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
+# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
+# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
+# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
+# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
+# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
+# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
+# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
+# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
+# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
+# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
+# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
+# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
+# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
+# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
+# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
+# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
+# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
+# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
+# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
+# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
+# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
+# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
+# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
+# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
+# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
+# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
+# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
+# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
+# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
+# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
+# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
+# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
+# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
+# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
+# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
+# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
+# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
+# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
+# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
+# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
+# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
+# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
+# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
+# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
+# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
+# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
+# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
+# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
+# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
+# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
+# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
+# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
+# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
+# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
+# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
+# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
+# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
+# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
+# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
+# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
+# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
+# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
+# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
+# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
+# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
+# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
+# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
+# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
+# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
+# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
+# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
+# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
+# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
+# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
+# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
+# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
+# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
+# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
+# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
+# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
+# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
+# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
+# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
+# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
+# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
+# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
+# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
+# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
+# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
+# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
+# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
+# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
+# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
+# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
+# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
+# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
+# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
+# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
+# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
+# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
+# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
+# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
+# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
+# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
+# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
+# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
+# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
+# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
+# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
+# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
+# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
+# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
+# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
+# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
+# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
+# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
+# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
+# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
+# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
+# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
+# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
+# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
+# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
+# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
+# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
+# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
+# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
+# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
+# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
+# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
+# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
+# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
+# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
+# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
+# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
+# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
+# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
+# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
+# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
+# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
+# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
+# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
+# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
+# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
+# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
+# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
+# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
+# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
+# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
+# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
+# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
+# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
+# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
+# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
+# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
+# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
+# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
+# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
+# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
+# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
+# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
+# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
+# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
+# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
+# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
+# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
+# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
+# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
+# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
+# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
+# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
+# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
+# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
+# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
+# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
+# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
+# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
+# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
+# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
+# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
+# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
+# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
+# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
+# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
+# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
+# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
+# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
+# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
+# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
+# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
+# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
+# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
+# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
+# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
+# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
+# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
+# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
+# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
+# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
+# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
+# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
+# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
+# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
+# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
+# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
+# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
+# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
+# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
+# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
+# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
+# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
+# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
+# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
+# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
+# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
+# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
+# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
+# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
+# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
+# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
+# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
+# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
+# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
+# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
+# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
+# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
+# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
+# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
+# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
+# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
+# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
+# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
+# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
+# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
+# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
+# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
+# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
+# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
+# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
+# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
+# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
+# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
+# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
+# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
+# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
+# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
+# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
+# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
+# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
+# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
+# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
+# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
+# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
+# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
+# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
+# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
+# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
+# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
+# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
+# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
+# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
+# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
+# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
+# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
+# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
+# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
+# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
+# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
+# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
+# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
+# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
+# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
+# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
+# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
+# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
+# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
+# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
+# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
+# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
+# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
+# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
+# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
+# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
+# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
+# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
+# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
+# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
+# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
+# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
+# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
+# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
+# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
+# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
+# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
+# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
+# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
+# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
+# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
+# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
+# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
+# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
+# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
+# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
+# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
+# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
+# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
+# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
+# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
+# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
+# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
+# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
+# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
+# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
+# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
+# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
+# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
+# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
+# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
+# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
+# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
+# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
+# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
+# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
+# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
+# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
+# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
+# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
+# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
+# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
+# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
+# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
+# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
+# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
+# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
+# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
+# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
+# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
+# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
+# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
+# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
+# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
+# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
+# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
+# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
+# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
+# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
+# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
+# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
+# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
+# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
+# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
+# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
+# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
+# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
+# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
+# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
+# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
+# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
+# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
+# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
+# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
+# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
+# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
+# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
+# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
+# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
+# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
+# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
+# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
+# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
+# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
+# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
+# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
+# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
+# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
+# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
+# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
+# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
+# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
+# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
+# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
+# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
+# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
+# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
+# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
+# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
+# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
+# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
+# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
+# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
+# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
+# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
+# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
+# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
+# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
+# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
+# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
+# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
+# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
+# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
+# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
+# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
+# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
+# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
+# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
+# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
+# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
+# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
+# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
+# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
+# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
+# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
+# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
+# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
+# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
+# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
+# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
+# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
+# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
+# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
+# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
+# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
+# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
+# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
+# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
+# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
+# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
+# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
+# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
+# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
+# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
+# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
+# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
+# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
+# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
+# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
+# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
+# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
+# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
+# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
+# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
+# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
+# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
+# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
+# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
+# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
+# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
+# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
+# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
+# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
+# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
+# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
+# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
+# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
+# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
+# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
+# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
+# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
+# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
+# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
+# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
+# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
+# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
+# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
+# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
+# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
+# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
+# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
+# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
+# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
+# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
+# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
+# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
+# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
+# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
+# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
+# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
+# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
+# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
+# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
+# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
+# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
+# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
+# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
+# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
+# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
+# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
+# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
+# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
+# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
+# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
+# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
+# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
+# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
+# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
+# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
+# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
+# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
+# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
+# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
+# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
+# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
+# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
+# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
+# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
+# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
+# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
+# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
+# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
+# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
+# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
+# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
+# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
+# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
+# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
+# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
+# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
+# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
+# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
+# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
+# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
+# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
+# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
+# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
+# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
+# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
+# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
+# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
+# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
+# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
+# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
+# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
+# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
+# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
+# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
+# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
+# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
+# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
+# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
+# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
+# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
+# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
+# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
+# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
+# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
+# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
+# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
+# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
+# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
+# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
+# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
+# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
+# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
+# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
+# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
+# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
+# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
+# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
+# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
+# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
+# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
+# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
+# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
+# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
+# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
+# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
+# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
+# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
+# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
+# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
+# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
+# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
+# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
+# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
+# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
+# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
+# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
+# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
+# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
+# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
+# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
+# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
+# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
+# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
+# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
+# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
+# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
+# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
+# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
+# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
+# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
+# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
+# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
+# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
+# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
+# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
+# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
+# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
+# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
+# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
+# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
+# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
+# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
+# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
+# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
+# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
+# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
+# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
+# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
+# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
+# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
+# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
+# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
+# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
+# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
+# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
+# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
+# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
+# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
+# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
+# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
+# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
+# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
+# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
+# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
+# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
+# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
+# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
+# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
+# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
+# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
+# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
+# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
+# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
+# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
+# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
+# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
+# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
+# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
+# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
+# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
+# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
+# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
+# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
+# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
+# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
+# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
+# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
+# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
+# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
+# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
+# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
+# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
+# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
+# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
+# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
+# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
+# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
+# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
+# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
+# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
+# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
+# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
+# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
+# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
+# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
+# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
+# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
+# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
+# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
+# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
+# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
+# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
+# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
+# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
+# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
+# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
+# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
+# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
+# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
+# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
+# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
+# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
+# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
+# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
+# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
+# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
+# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
+# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
+# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
+# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
+# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
+# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
+# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
+# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
+# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
+# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
+# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
+# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
+# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
+# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
+# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
+# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
+# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
+# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
+# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
+# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
+# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
+# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
+# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
+# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
+# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
+# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
+# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
+# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
+# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
+# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
+# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
+# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
+# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
+# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
+# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
+# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
+# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
+# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
+# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
+# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
+# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
+# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
+# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
+# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
+# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
+# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
+# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
+# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
+# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
+# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
+# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
+# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
+# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
+# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
+# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
+# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
+# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
+# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
+# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
+# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
+# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
+# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
+# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
+# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
+# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
+# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
+# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
+# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
+# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
+# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
+# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
+# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
+# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
+# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
+# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
+# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
+# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
+# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
+# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
+# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
+# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
+# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
+# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
+# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
+# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
+# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
+# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
+# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
+# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
+# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
+# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
+# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
+# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
+# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
+# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
+# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
+# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
+# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
+# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
+# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
+# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
+# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
+# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
+# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
+# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
+# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
+# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
+# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
+# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
+# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
+# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
+# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
+# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
+# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
+# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
+# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
+# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
+# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
+# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
+# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
+# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
+# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
+# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
+# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
+# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
+# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
+# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
+# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
+# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
+# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
+# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
+# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
+# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
+# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
+# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
+# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
+# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
+# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
+# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
+# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
+# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
+# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
+# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
+# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
+# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
+# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
+# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
+# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
+# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
+# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
+# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
+# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
+# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
+# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
+# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
+# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
+# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
+# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
+# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
+# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
+# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
+# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
+# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
+# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
+# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
+# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
+# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
+# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
+# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
+# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
+# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
+# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
+# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
+# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
+# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
+# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
+# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
+# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
+# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
+# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
+# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
+# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
+# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
+# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
+# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
+# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
+# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
+# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
+# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
+# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
+# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
+# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
+# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
+# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
+# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
+# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
+# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
+# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
+# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
+# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
+# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
+# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
+# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
+# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
+# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
+# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
+# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
+# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
+# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
+# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
+# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
+# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
+# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
+# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
+# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
+# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
+# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
+# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
+# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
+# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
+# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
+# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
+# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
+# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
+# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
+# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
+# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
+# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
+# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
+# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
+# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
+# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
+# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
+# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
+# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
+# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
+# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
+# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
+# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
+# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
+# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
+# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
+# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
+# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
+# 11:38:37 --- INFO: Training process end.
+# 11:38:37 --- INFO: Evaluating process start.
+# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
+# 11:38:37 --- INFO: Evaluating process end.
+# 11:38:37 --- INFO: Predicting process start.
+# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
+# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
+# 11:38:38 --- INFO: Predicting process end.
+# 11:38:38 --- INFO: --------------------- GNN end ----------------------
+# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
+# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
+# 11:38:38 --- INFO: Loading dataset.
+# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:38:39 --- INFO: Calculating NTScore for each niche.
+# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
+# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
+# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
+# 11:38:39 --- INFO: Output NTScore tables.
+# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
16.1.3 visualization
16.1.3.1 load ONTraC results
-
+
The NTScore and binarized niche cluster info were stored in cell metadata
-
-# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
-# <char> <char> <char> <char> <char> <char> <char> <int> <num>
-# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
-# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
-# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
-# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
-# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
-# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
+
+# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
+# <char> <char> <char> <char> <char> <char> <char> <int> <num>
+# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
+# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
+# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
+# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
+# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
+# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
The probability matrix of each cell assigned to each niche cluster and
connectivity between niche cluster were stored here.
-
-
+
+
16.1.3.2 niche cluster prob
-spatFeatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'niche cluster',
- expression_values = "prob",
- group_by = 'list_ID',
- feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
- point_border_col = 'gray'
- )
+spatFeatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'niche cluster',
+ expression_values = "prob",
+ group_by = 'list_ID',
+ feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
+ point_border_col = 'gray'
+ )
16.1.3.3 binarized niche cluster
-spatPlot2D(giotto_obj,
- spat_unit = "cell",
- group_by = 'list_ID',
- cell_color = "NicheCluster",
- color_as_factor = T,
- point_size = 1,
- point_border_stroke = NA,
- title="Niche cluster")
+spatPlot2D(giotto_obj,
+ spat_unit = "cell",
+ group_by = 'list_ID',
+ cell_color = "NicheCluster",
+ color_as_factor = T,
+ point_size = 1,
+ point_border_stroke = NA,
+ title="Niche cluster")
16.1.3.5 NT score
-spatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'rna',
- group_by = 'list_ID',
- cell_color = "NTScore",
- color_as_factor = FALSE,
- cell_color_gradient = "turbo",
- point_size = 1,
- point_border_stroke = NA
- )
+spatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ group_by = 'list_ID',
+ cell_color = "NTScore",
+ color_as_factor = FALSE,
+ cell_color_gradient = "turbo",
+ point_size = 1,
+ point_border_stroke = NA
+ )
-
+
diff --git a/interoperability-with-other-frameworks.html b/interoperability-with-other-frameworks.html
index 7050533..347d3af 100644
--- a/interoperability-with-other-frameworks.html
+++ b/interoperability-with-other-frameworks.html
@@ -521,10 +521,10 @@ 15 Interoperability with other fr
15.1 Load Giotto object
To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process:
-
+
Load a Giotto mini Visium object, which will be used for demonstrating interoperability.
-
+
15.2 Seurat
@@ -533,99 +533,99 @@ 15.2 Seurat
15.2.1 Conversion of Giotto Object to Seurat Object
To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function
-
+
Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data.
-gToS <- giottoToSeuratV5(gobject = gobject,
- spat_unit = "cell")
-
-plot1 <- VlnPlot(gToS,
- features = "nCount_rna",
- pt.size = 0.1) +
- NoLegend()
-
-plot2 <- SpatialFeaturePlot(gToS,
- features = "nCount_rna",
- pt.size.factor = 2) +
- theme(legend.position = "right")
-
-wrap_plots(plot1, plot2)
+gToS <- giottoToSeuratV5(gobject = gobject,
+ spat_unit = "cell")
+
+plot1 <- VlnPlot(gToS,
+ features = "nCount_rna",
+ pt.size = 0.1) +
+ NoLegend()
+
+plot2 <- SpatialFeaturePlot(gToS,
+ features = "nCount_rna",
+ pt.size.factor = 2) +
+ theme(legend.position = "right")
+
+wrap_plots(plot1, plot2)
15.2.1.1 Apply SCTransform
We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function.
-
+
15.2.1.2 Dimension Reduction
We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object:
-
+
15.2.2 Conversion of Seurat object Back to Giotto Object
To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks.
-giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
- spatial_assay = "rna",
- dim_reduction = c("pca", "umap"),
- sp_network = "Delaunay_network",
- nn_network = c("sNN.pca", "custom_NN" ))
+giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
+ spatial_assay = "rna",
+ dim_reduction = c("pca", "umap"),
+ sp_network = "Delaunay_network",
+ nn_network = c("sNN.pca", "custom_NN" ))
15.2.2.1 Clustering and Plotting UMAP
Here we perform K-means clustering on the UMAP results obtained from the Seurat object:
-## k-means clustering
-giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
- dim_reduction_to_use = "pca")
-
-#Plot UMAP post-clustering to visualize kmeans
-graph2 <- Giotto::plotUMAP(
- gobject = giottoFromSeurat,
- cell_color = "kmeans",
- show_NN_network = TRUE,
- point_size = 2.5
-)
+## k-means clustering
+giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
+ dim_reduction_to_use = "pca")
+
+#Plot UMAP post-clustering to visualize kmeans
+graph2 <- Giotto::plotUMAP(
+ gobject = giottoFromSeurat,
+ cell_color = "kmeans",
+ show_NN_network = TRUE,
+ point_size = 2.5
+)
15.2.2.2 Spatial CoExpression
We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function:
-ranktest <- binSpect(giottoFromSeurat,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "Delaunay_network")
-
-ext_spatial_genes <- ranktest[1:300,]$feats
-
-spat_cor_netw_DT <- detectSpatialCorFeats(
- giottoFromSeurat,
- method = "network",
- spatial_network_name = "Delaunay_network",
- subset_feats = ext_spatial_genes)
-
-top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
- feats = "Dsp",
- show_top_feats = 10)
-
-spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
- name = "spat_netw_clus",
- k = 7)
-heatmSpatialCorFeats(
- giottoFromSeurat,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL),
- save_plot = TRUE,
- show_plot = TRUE,
- return_plot = FALSE,
- save_param = list(base_height = 6, base_width = 8, units = 'cm'))
+ranktest <- binSpect(giottoFromSeurat,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "Delaunay_network")
+
+ext_spatial_genes <- ranktest[1:300,]$feats
+
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ giottoFromSeurat,
+ method = "network",
+ spatial_network_name = "Delaunay_network",
+ subset_feats = ext_spatial_genes)
+
+top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
+ feats = "Dsp",
+ show_top_feats = 10)
+
+spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
+ name = "spat_netw_clus",
+ k = 7)
+heatmSpatialCorFeats(
+ giottoFromSeurat,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ return_plot = FALSE,
+ save_param = list(base_height = 6, base_width = 8, units = 'cm'))
@@ -635,60 +635,60 @@ 15.3 SpatialExperiment
To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries.
-library(SpatialExperiment)
-library(ggspavis)
-library(pheatmap)
-library(scater)
-library(scran)
-library(nnSVG)
+library(SpatialExperiment)
+library(ggspavis)
+library(pheatmap)
+library(scater)
+library(scran)
+library(nnSVG)
15.3.1 Convert Giotto Object to SpatialExperiment Object
To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function.
-
+
The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use:
-
+
15.3.1.1 Identify top spatially variable genes with nnSVG
We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate:
-# One of the assays should be "logcounts"
-# We rename the normalized assay to "logcounts"
-assayNames(spe)[[2]] <- "logcounts"
-
-# Create model matrix for leiden clustering labels
-X <- model.matrix(~ colData(spe)$leiden_clus)
-dim(X)
+# One of the assays should be "logcounts"
+# We rename the normalized assay to "logcounts"
+assayNames(spe)[[2]] <- "logcounts"
+
+# Create model matrix for leiden clustering labels
+X <- model.matrix(~ colData(spe)$leiden_clus)
+dim(X)
- Run nnSVG
This step will take several minutes to run
-
+
15.3.2 Conversion of SpatialExperiment object back to Giotto
We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object.
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-print(giottoFromSPE)
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+print(giottoFromSPE)
15.3.2.1 Plotting top genes from nnSVG in Giotto
Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object.
-ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
- output = "data.table")
-
-ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
-spatFeatPlot2D(giottoFromSPE,
- expression_values = "scaled_rna_cell",
- feats = ext_spatial_genes[1:4],
- point_size = 2)
+ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
+ output = "data.table")
+
+ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
+spatFeatPlot2D(giottoFromSPE,
+ expression_values = "scaled_rna_cell",
+ feats = ext_spatial_genes[1:4],
+ point_size = 2)
@@ -701,97 +701,97 @@ 15.4 AnnData
15.4.1 Load Required Libraries
To start, we need to load the necessary libraries, including reticulate for interfacing with Python.
-
+
15.4.2 Specify Path for Results
First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions.
-# Specify path to which results may be saved
-results_directory <- "results/03_session4/giotto_anndata_conversion/"
-
-instrs <- showGiottoInstructions(gobject)
-
-mini_gobject <- replaceGiottoInstructions(gobject = gobject,
- instructions = instrs)
+# Specify path to which results may be saved
+results_directory <- "results/03_session4/giotto_anndata_conversion/"
+
+instrs <- showGiottoInstructions(gobject)
+
+mini_gobject <- replaceGiottoInstructions(gobject = gobject,
+ instructions = instrs)
15.4.2.1 Create Default kNN Network
We will create a k-nearest neighbor (kNN) network using mostly default parameters.
-
+
15.4.3 Giotto To AnnData
To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData()
-
+
Next, we import scanpy and perform a series of preprocessing steps on the AnnData object.
-scanpy <- import("scanpy")
-
-adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
-
-# Normalize total counts per cell
-scanpy$pp$normalize_total(adata, target_sum=1e4)
-
-# Log-transform the data
-scanpy$pp$log1p(adata)
-
-# Perform PCA
-scanpy$pp$pca(adata, n_comps=40L)
-
-# Compute the neighborhood graph
-scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
-
-# Run UMAP
-scanpy$tl$umap(adata)
-
-# Save the processed AnnData object
-adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
-
-processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
+scanpy <- import("scanpy")
+
+adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
+
+# Normalize total counts per cell
+scanpy$pp$normalize_total(adata, target_sum=1e4)
+
+# Log-transform the data
+scanpy$pp$log1p(adata)
+
+# Perform PCA
+scanpy$pp$pca(adata, n_comps=40L)
+
+# Compute the neighborhood graph
+scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
+
+# Run UMAP
+scanpy$tl$umap(adata)
+
+# Save the processed AnnData object
+adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
+
+processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
15.4.4 Convert AnnData to Giotto
Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto.
-
+
15.4.4.1 UMAP Visualization
Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object.
-Giotto::plotUMAP(
- gobject = giottoFromAnndata,
- dim_reduction_name = "umap.ad",
- cell_color = "leiden_clus",
- point_size = 3
-)
+Giotto::plotUMAP(
+ gobject = giottoFromAnndata,
+ dim_reduction_name = "umap.ad",
+ cell_color = "leiden_clus",
+ point_size = 3
+)
15.5 Create mini Vizgen object
-mini_gobject <- loadGiottoMini(dataset = "vizgen",
- python_path = NULL)
-
-mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
- instructions = instrs)
-mini_gobject <- createNearestNetwork(gobject = mini_gobject,
- spat_unit = "aggregate",
- feat_type = "rna",
- type = "kNN",
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap",
- k = 6,
- name = "new_network")
+mini_gobject <- loadGiottoMini(dataset = "vizgen",
+ python_path = NULL)
+
+mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
+ instructions = instrs)
+mini_gobject <- createNearestNetwork(gobject = mini_gobject,
+ spat_unit = "aggregate",
+ feat_type = "rna",
+ type = "kNN",
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap",
+ k = 6,
+ name = "new_network")
Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details.
-
+
diff --git a/introduction-to-the-giotto-package.html b/introduction-to-the-giotto-package.html
index e22d6d1..0ab2014 100644
--- a/introduction-to-the-giotto-package.html
+++ b/introduction-to-the-giotto-package.html
@@ -546,19 +546,58 @@ 4.3 Installation + python environ
4.3.1 Giotto installation
Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository.
Much of this already covered in Section 2.2, but the highlights are:
-
-- Ensure your system
-
-To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak.
+
+4.3.1.1 System prerequisites
+
+- for windows, Rtools needs to be installed
+- a major dependency terra needs
GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux
+
+
+
+4.3.1.2 Installation of released version
+To install the currently released version of Giotto in a single step:
+This should automatically install all the Giotto dependencies and other Giotto module packages.
+
+
+4.3.1.3 Installation of dev branch Giotto packages
+You can install dev branch versions by using devtools::install_github()
+Core module dev branchs:
+
+- “drieslab/Giotto@suite_dev”
+- “drieslab/GiottoVisuals@dev”
+- “drieslab/GiottoClass@dev”
+- “drieslab/GiottoUtils@dev”
+
+
+pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages.
+
+
+4.3.1.4 Common install issues
+If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work.
+
+If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it.
+
+## # Downloading packages -------------------------------------------------------
+## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s]
+## Successfully downloaded 1 package in 1.2 seconds.
+##
+## The following package(s) will be installed:
+## - irlba [2.3.5.1]
+## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu".
+##
+## # Installing packages --------------------------------------------------------
+## - Installing irlba ... OK [built from source and cached in 5.7s]
+## Successfully installed 1 package in 5.7 seconds.
+
4.3.2 Python environment
4.3.2.1 Default installation
In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment.
-
+
Two things are needed for python to work:
- A conda (e.g. miniconda or anaconda) installation which is the package and environment management system.
@@ -580,17 +619,17 @@ 4.3.2.1 Default installation
4.3.2.2 Custom installs
Custom python environments can be made by first setting up a new environment and establishing the name and python version to use.
-
+
Following that, one or more python packages to install can be added to the environment.
-reticulate::py_install(
- pip = TRUE,
- envname = '[name of env]',
- packages = c(
- "package1",
- "package2",
- "..."
- )
-)
+reticulate::py_install(
+ pip = TRUE,
+ envname = '[name of env]',
+ packages = c(
+ "package1",
+ "package2",
+ "..."
+ )
+)
Once an environment has been set up, Giotto can hook into it.
@@ -612,17 +651,56 @@ 4.3.2.3 Using a specific environm
Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto.
You would run file.edit("~/.Rprofile") and then add options("giotto.py_path" = "[path to env or envname]") as a line so that it is automatically set at the start of each session.
If a specific environment should only be used a couple times then method 1 is easiest:
-
+
To check which conda environments exist on your machine:
-
+
Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running:
-
+
4.4 Giotto instructions
-Giotto
+Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are:
+
+- python_path - when set, will activate a python environment
+- save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render.
+- save_plot - whether to save plots to the
save_dir
+- return_plot - whether to return the plot objects. When FALSE, only NULL is returned
+- show_plot - whether to show the plot in the viewer
+
+These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function.
+library(Giotto)
+save_dir <- "results/01_session2/"
+
+# this call will also intialize the python env
+instrs <- createGiottoInstructions(
+ save_dir = save_dir, # working directory is the default
+ show_plot = FALSE,
+ save_plot = TRUE,
+ return_plot = FALSE,
+ python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default
+)
+force(instrs)
+Giotto object creation functions all have an instructions param for passing in instructions objects.
+giotto objects will also respond to the instructions() generic.
+test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object
+
+# example plot
+g <- GiottoData::loadGiottoMini("visium")
+instructions(g) <- instrs
+instructions(g, "show_plot") # instructions say not to plot to viewer
+spatPlot2D(g, show_image = TRUE, image_name = "image")
+# instead it will directly write to the results folder
+As an example, you can also set individual instructions
+
+
+
+
+Figure 4.2: example image output
+
+
diff --git a/multi-omics-integration.html b/multi-omics-integration.html
index 47cad23..e76e25a 100644
--- a/multi-omics-integration.html
+++ b/multi-omics-integration.html
@@ -520,14 +520,14 @@ 14 Multi-omics integration
14.1 The CytAssist technology
The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image.
-
+
Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics.
The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument.
-
+
Figure 14.2: CytAssist workflow. Source: 10X genomics.
@@ -542,19 +542,19 @@
14.2 Introduction to the spatial
14.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/03_session3")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
+dir.create("data/03_session3")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”.
-
+
14.4 Create the Giotto object
@@ -564,124 +564,124 @@ 14.4 Create the Giotto objectx,y(,z) coordinates for cells or spots (or the path to)
createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/03_session3/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-# Provide the path to the data folder
-data_path <- "data/03_session3/"
-
-# Create object directly from the data folder
-visium_tonsil <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/03_session3/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+# Provide the path to the data folder
+data_path <- "data/03_session3/"
+
+# Create object directly from the data folder
+visium_tonsil <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
- Print the information of the object, note that both rna and protein are listed in the expression slot.
-
+
14.5 Subset on spots that were covered by tissue
-spatPlot2D(
- gobject = visium_tonsil,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE,
- image_name = "image"
-)
-
+spatPlot2D(
+ gobject = visium_tonsil,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE,
+ image_name = "image"
+)
+
Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object.
-
+
14.6 RNA processing
- Run the Filtering, normalization, and statistics steps using only the RNA feature.
-visium_tonsil <- filterGiotto(
- gobject = visium_tonsil,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil)
+visium_tonsil <- filterGiotto(
+ gobject = visium_tonsil,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1000,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil)
- Dimension reduction
Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs.
-visium_tonsil <- calculateHVF(gobject = visium_tonsil)
-
-visium_tonsil <- runPCA(gobject = visium_tonsil)
+visium_tonsil <- calculateHVF(gobject = visium_tonsil)
+
+visium_tonsil <- runPCA(gobject = visium_tonsil)
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the RNA-based Leiden clusters.
-
+
- Visualization
Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot.
-plotUMAP(gobject = visium_tonsil,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters.
Plot the spatial distribution of the RNA-based Leiden cluster per spot.
-spatPlot2D(gobject = visium_tonsil,
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.5: Spatial distribution of RNA-based Leiden clusters.
@@ -693,80 +693,80 @@
14.7 Protein processingvisium_tonsil <- filterGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein")
+visium_tonsil <- filterGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein")
- Dimension reduction
Calculate the principal components using all the proteins available in the dataset.
-
+
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the Protein-based Leiden clusters.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- resolution = 1,
- n_iterations = 1000)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ resolution = 1,
+ n_iterations = 1000)
- Visualization
Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters.
Plot the spatial distribution of the Protein-based Leiden clusters.
-spatPlot2D(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.7: Spatial distribution of Protein-based Leiden clusters.
@@ -779,68 +779,68 @@
14.8 Multi-omics integrationCalculate kNN
Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type.
-## RNA modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
-
-## Protein modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
+## RNA modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
+
+## Protein modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
- Run WNN
Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object.
-visium_tonsil <- runWNN(visium_tonsil,
- spat_unit = "cell",
- modality_1 = "rna",
- modality_2 = "protein",
- pca_name_modality_1 = "pca",
- pca_name_modality_2 = "protein.pca",
- k = 20,
- integrated_feat_type = NULL,
- matrix_result_name = NULL,
- w_name_modality_1 = NULL,
- w_name_modality_2 = NULL,
- verbose = TRUE)
+visium_tonsil <- runWNN(visium_tonsil,
+ spat_unit = "cell",
+ modality_1 = "rna",
+ modality_2 = "protein",
+ pca_name_modality_1 = "pca",
+ pca_name_modality_2 = "protein.pca",
+ k = 20,
+ integrated_feat_type = NULL,
+ matrix_result_name = NULL,
+ w_name_modality_1 = NULL,
+ w_name_modality_2 = NULL,
+ verbose = TRUE)
- Run Integrated umap
Calculate the UMAP using the weights of each feature per spot.
-visium_tonsil <- runIntegratedUMAP(visium_tonsil,
- modality1 = "rna",
- modality2 = "protein",
- spread = 7,
- min_dist = 1,
- force = FALSE)
+visium_tonsil <- runIntegratedUMAP(visium_tonsil,
+ modality1 = "rna",
+ modality2 = "protein",
+ spread = 7,
+ min_dist = 1,
+ force = FALSE)
- Calculate integrated clusters
Calculate the multiomics-based Leiden clusters using the weights of each feature per spot.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- nn_network_to_use = "kNN",
- network_name = "integrated_kNN",
- name = "integrated_leiden_clus",
- resolution = 1)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ nn_network_to_use = "kNN",
+ network_name = "integrated_kNN",
+ name = "integrated_leiden_clus",
+ resolution = 1)
- Visualize the integrated UMAP
Plot the integrated UMAP and color the spots using the integrated Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- dim_reduction_name = "integrated.umap",
- point_size = 2,
- title = "Integrated UMAP using Integrated Leiden clusters")
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ dim_reduction_name = "integrated.umap",
+ point_size = 2,
+ title = "Integrated UMAP using Integrated Leiden clusters")
+
Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters.
@@ -850,14 +850,14 @@
14.8 Multi-omics integrationVisualize spatial plot with integrated clusters
Plot the spatial distribution of the integrated Leiden clusters.
-spatPlot2D(visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- point_size = 3,
- show_image = TRUE,
- title = "Integrated Leiden clustering")
-
+spatPlot2D(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ point_size = 3,
+ show_image = TRUE,
+ title = "Integrated Leiden clustering")
+
Figure 14.9: Spatial distribution of the integrated Leiden clusters.
@@ -866,85 +866,85 @@
14.8 Multi-omics integration
14.9 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
- [7] matrixStats_1.3.0 compiler_4.4.1
- [9] png_0.1-8 systemfonts_1.1.0
- [11] vctrs_0.6.5 reshape2_1.4.4
- [13] stringr_1.5.1 pkgconfig_2.0.3
- [15] SpatialExperiment_1.14.0 crayon_1.5.3
- [17] fastmap_1.2.0 backports_1.5.0
- [19] magick_2.8.4 XVector_0.44.0
- [21] labeling_0.4.3 utf8_1.2.4
- [23] rmarkdown_2.27 UCSC.utils_1.0.0
- [25] ragg_1.3.2 purrr_1.0.2
- [27] xfun_0.46 beachmat_2.20.0
- [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
- [31] jsonlite_1.8.8 DelayedArray_0.30.1
- [33] BiocParallel_1.38.0 terra_1.7-78
- [35] irlba_2.3.5.1 parallel_4.4.1
- [37] R6_2.5.1 stringi_1.8.4
- [39] RColorBrewer_1.1-3 reticulate_1.38.0
- [41] parallelly_1.37.1 GenomicRanges_1.56.1
- [43] scattermore_1.2 Rcpp_1.0.13
- [45] bookdown_0.40 SummarizedExperiment_1.34.0
- [47] knitr_1.48 future.apply_1.11.2
- [49] R.utils_2.12.3 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
+ [7] matrixStats_1.3.0 compiler_4.4.1
+ [9] png_0.1-8 systemfonts_1.1.0
+ [11] vctrs_0.6.5 reshape2_1.4.4
+ [13] stringr_1.5.1 pkgconfig_2.0.3
+ [15] SpatialExperiment_1.14.0 crayon_1.5.3
+ [17] fastmap_1.2.0 backports_1.5.0
+ [19] magick_2.8.4 XVector_0.44.0
+ [21] labeling_0.4.3 utf8_1.2.4
+ [23] rmarkdown_2.27 UCSC.utils_1.0.0
+ [25] ragg_1.3.2 purrr_1.0.2
+ [27] xfun_0.46 beachmat_2.20.0
+ [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
+ [31] jsonlite_1.8.8 DelayedArray_0.30.1
+ [33] BiocParallel_1.38.0 terra_1.7-78
+ [35] irlba_2.3.5.1 parallel_4.4.1
+ [37] R6_2.5.1 stringi_1.8.4
+ [39] RColorBrewer_1.1-3 reticulate_1.38.0
+ [41] parallelly_1.37.1 GenomicRanges_1.56.1
+ [43] scattermore_1.2 Rcpp_1.0.13
+ [45] bookdown_0.40 SummarizedExperiment_1.34.0
+ [47] knitr_1.48 future.apply_1.11.2
+ [49] R.utils_2.12.3 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/reference-keys.txt b/reference-keys.txt
index a17604d..555f956 100644
--- a/reference-keys.txt
+++ b/reference-keys.txt
@@ -4,115 +4,116 @@ fig:unnamed-chunk-14
fig:unnamed-chunk-15
fig:unnamed-chunk-16
fig:unnamed-chunk-18
-fig:unnamed-chunk-30
-fig:unnamed-chunk-35
-fig:unnamed-chunk-38
-fig:unnamed-chunk-40
+fig:unnamed-chunk-32
+fig:unnamed-chunk-37
fig:unnamed-chunk-42
-fig:unnamed-chunk-44
+fig:unnamed-chunk-45
+fig:unnamed-chunk-47
fig:unnamed-chunk-49
fig:unnamed-chunk-51
-fig:unnamed-chunk-53
-fig:unnamed-chunk-55
-fig:unnamed-chunk-59
-fig:unnamed-chunk-61
-fig:unnamed-chunk-63
+fig:unnamed-chunk-56
+fig:unnamed-chunk-58
+fig:unnamed-chunk-60
+fig:unnamed-chunk-62
fig:unnamed-chunk-66
-fig:unnamed-chunk-69
-fig:unnamed-chunk-74
+fig:unnamed-chunk-68
+fig:unnamed-chunk-70
+fig:unnamed-chunk-73
fig:unnamed-chunk-76
-fig:unnamed-chunk-78
-fig:unnamed-chunk-80
-fig:unnamed-chunk-82
-fig:unnamed-chunk-84
+fig:unnamed-chunk-81
+fig:unnamed-chunk-83
+fig:unnamed-chunk-85
fig:unnamed-chunk-87
+fig:unnamed-chunk-89
+fig:unnamed-chunk-91
fig:unnamed-chunk-94
-fig:unnamed-chunk-96
-fig:unnamed-chunk-98
fig:unnamed-chunk-101
fig:unnamed-chunk-103
fig:unnamed-chunk-105
+fig:unnamed-chunk-108
+fig:unnamed-chunk-110
fig:unnamed-chunk-112
-fig:unnamed-chunk-114
-fig:unnamed-chunk-118
-fig:unnamed-chunk-120
-fig:unnamed-chunk-123
-fig:unnamed-chunk-128
-fig:unnamed-chunk-131
-fig:unnamed-chunk-134
-fig:unnamed-chunk-137
+fig:unnamed-chunk-119
+fig:unnamed-chunk-121
+fig:unnamed-chunk-125
+fig:unnamed-chunk-127
+fig:unnamed-chunk-130
+fig:unnamed-chunk-135
+fig:unnamed-chunk-138
fig:unnamed-chunk-141
-fig:unnamed-chunk-143
-fig:unnamed-chunk-147
+fig:unnamed-chunk-144
+fig:unnamed-chunk-148
fig:unnamed-chunk-150
-fig:unnamed-chunk-152
+fig:unnamed-chunk-154
+fig:unnamed-chunk-157
fig:unnamed-chunk-159
-fig:unnamed-chunk-161
-fig:unnamed-chunk-163
-fig:unnamed-chunk-165
-fig:unnamed-chunk-167
-fig:unnamed-chunk-169
-fig:unnamed-chunk-171
+fig:unnamed-chunk-166
+fig:unnamed-chunk-168
+fig:unnamed-chunk-170
+fig:unnamed-chunk-172
fig:unnamed-chunk-174
-fig:unnamed-chunk-175
fig:unnamed-chunk-176
-fig:unnamed-chunk-185
-fig:unnamed-chunk-191
-fig:unnamed-chunk-194
+fig:unnamed-chunk-178
+fig:unnamed-chunk-181
+fig:unnamed-chunk-182
+fig:unnamed-chunk-183
+fig:unnamed-chunk-192
+fig:unnamed-chunk-198
fig:unnamed-chunk-201
-fig:unnamed-chunk-204
-fig:unnamed-chunk-206
-fig:unnamed-chunk-210
-fig:unnamed-chunk-212
-fig:unnamed-chunk-215
+fig:unnamed-chunk-208
+fig:unnamed-chunk-211
+fig:unnamed-chunk-213
fig:unnamed-chunk-217
fig:unnamed-chunk-219
-fig:unnamed-chunk-221
+fig:unnamed-chunk-222
fig:unnamed-chunk-224
-fig:unnamed-chunk-225
-fig:unnamed-chunk-227
-fig:unnamed-chunk-229
-fig:unnamed-chunk-230
+fig:unnamed-chunk-226
+fig:unnamed-chunk-228
+fig:unnamed-chunk-231
fig:unnamed-chunk-232
fig:unnamed-chunk-234
fig:unnamed-chunk-236
-fig:unnamed-chunk-275
-fig:unnamed-chunk-279
-fig:unnamed-chunk-281
-fig:unnamed-chunk-283
+fig:unnamed-chunk-237
+fig:unnamed-chunk-239
+fig:unnamed-chunk-241
+fig:unnamed-chunk-243
+fig:unnamed-chunk-282
fig:unnamed-chunk-286
fig:unnamed-chunk-288
fig:unnamed-chunk-290
-fig:unnamed-chunk-292
-fig:unnamed-chunk-294
+fig:unnamed-chunk-293
+fig:unnamed-chunk-295
fig:unnamed-chunk-297
fig:unnamed-chunk-299
fig:unnamed-chunk-301
-fig:unnamed-chunk-303
-fig:unnamed-chunk-305
-fig:unnamed-chunk-365
-fig:unnamed-chunk-366
+fig:unnamed-chunk-304
+fig:unnamed-chunk-306
+fig:unnamed-chunk-308
+fig:unnamed-chunk-310
+fig:unnamed-chunk-312
fig:unnamed-chunk-372
+fig:unnamed-chunk-373
fig:unnamed-chunk-379
-fig:unnamed-chunk-381
-fig:unnamed-chunk-387
-fig:unnamed-chunk-389
-fig:unnamed-chunk-395
-fig:unnamed-chunk-397
-fig:unnamed-chunk-462
-fig:unnamed-chunk-465
+fig:unnamed-chunk-386
+fig:unnamed-chunk-388
+fig:unnamed-chunk-394
+fig:unnamed-chunk-396
+fig:unnamed-chunk-402
+fig:unnamed-chunk-404
fig:unnamed-chunk-469
-fig:unnamed-chunk-470
fig:unnamed-chunk-472
-fig:unnamed-chunk-474
fig:unnamed-chunk-476
-fig:unnamed-chunk-478
+fig:unnamed-chunk-477
fig:unnamed-chunk-479
fig:unnamed-chunk-481
+fig:unnamed-chunk-483
fig:unnamed-chunk-485
+fig:unnamed-chunk-486
fig:unnamed-chunk-488
-fig:unnamed-chunk-490
fig:unnamed-chunk-492
+fig:unnamed-chunk-495
+fig:unnamed-chunk-497
+fig:unnamed-chunk-499
giotto-suite-workshop-2024
instructors
topics-and-schedule
@@ -153,6 +154,10 @@ presentation-1
ecosystem
installation-python-environment
giotto-installation-1
+system-prerequisites
+installation-of-released-version
+installation-of-dev-branch-giotto-packages
+common-install-issues
python-environment
default-installation
custom-installs
diff --git a/search_index.json b/search_index.json
index 2736898..72ab7c3 100644
--- a/search_index.json
+++ b/search_index.json
@@ -1 +1 @@
-[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: Ensure your system To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak. pak::pak("drieslab/Giotto") 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
+[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: 4.3.1.1 System prerequisites for windows, Rtools needs to be installed a major dependency terra needs GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux 4.3.1.2 Installation of released version To install the currently released version of Giotto in a single step: pak::pak("drieslab/Giotto") This should automatically install all the Giotto dependencies and other Giotto module packages. 4.3.1.3 Installation of dev branch Giotto packages You can install dev branch versions by using devtools::install_github() Core module dev branchs: “drieslab/Giotto@suite_dev” “drieslab/GiottoVisuals@dev” “drieslab/GiottoClass@dev” “drieslab/GiottoUtils@dev” devtools::install_github("drieslab/GiottoClass@dev") pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages. 4.3.1.4 Common install issues If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work. devtools::install_version("Matrix", version = "1.6-5") If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it. install.packages("irlba", type = "source") ## # Downloading packages ------------------------------------------------------- ## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s] ## Successfully downloaded 1 package in 1.2 seconds. ## ## The following package(s) will be installed: ## - irlba [2.3.5.1] ## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu". ## ## # Installing packages -------------------------------------------------------- ## - Installing irlba ... OK [built from source and cached in 5.7s] ## Successfully installed 1 package in 5.7 seconds. 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are: python_path - when set, will activate a python environment save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render. save_plot - whether to save plots to the save_dir return_plot - whether to return the plot objects. When FALSE, only NULL is returned show_plot - whether to show the plot in the viewer These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function. library(Giotto) save_dir <- "results/01_session2/" # this call will also intialize the python env instrs <- createGiottoInstructions( save_dir = save_dir, # working directory is the default show_plot = FALSE, save_plot = TRUE, return_plot = FALSE, python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default ) force(instrs) Giotto object creation functions all have an instructions param for passing in instructions objects. giotto objects will also respond to the instructions() generic. test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object # example plot g <- GiottoData::loadGiottoMini("visium") instructions(g) <- instrs instructions(g, "show_plot") # instructions say not to plot to viewer spatPlot2D(g, show_image = TRUE, image_name = "image") # instead it will directly write to the results folder As an example, you can also set individual instructions instructions(g, "show_plot") <- TRUE spatPlot2D(g, show_image = TRUE, image_name = "image") Figure 4.2: example image output "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
diff --git a/spatial-multi-modal-analysis.html b/spatial-multi-modal-analysis.html
index 5bdab74..c78f952 100644
--- a/spatial-multi-modal-analysis.html
+++ b/spatial-multi-modal-analysis.html
@@ -563,48 +563,48 @@ 13.2.3 Spatial classes:# load in data
-library(Giotto)
-
-g <- GiottoData::loadGiottoMini("vizgen")
-
-activeSpatUnit(g) <- "aggregate"
-
-gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE)
-
-gimg <- getGiottoImage(g)
+
13.3 Examples of the simple transforms with a giottoPolygon
-rain <- rainbow(nrow(gpoly))
-
-line_width <- 0.3
-
-# par to setup the grid plotting layout
-p <- par(no.readonly = TRUE)
-par(mfrow=c(3,3))
-gpoly |>
- plot(main = "no transform", col = rain, lwd = line_width)
-
-gpoly |> spatShift(dx = 1000) |>
- plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
-
-gpoly |> spin(45) |>
- plot(main = "spin(45)", col = rain, lwd = line_width)
-
-gpoly |> rescale(fx = 10, fy = 6) |>
- plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
-
-gpoly |> flip(direction = "vertical") |>
- plot(main = "flip()", col = rain, lwd = line_width)
-
-gpoly |> t() |>
- plot(main = "t()", col = rain, lwd = line_width)
-
-gpoly |> shear(fx = 0.5) |>
- plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
-par(p)
+rain <- rainbow(nrow(gpoly))
+
+line_width <- 0.3
+
+# par to setup the grid plotting layout
+p <- par(no.readonly = TRUE)
+par(mfrow=c(3,3))
+gpoly |>
+ plot(main = "no transform", col = rain, lwd = line_width)
+
+gpoly |> spatShift(dx = 1000) |>
+ plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
+
+gpoly |> spin(45) |>
+ plot(main = "spin(45)", col = rain, lwd = line_width)
+
+gpoly |> rescale(fx = 10, fy = 6) |>
+ plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
+
+gpoly |> flip(direction = "vertical") |>
+ plot(main = "flip()", col = rain, lwd = line_width)
+
+gpoly |> t() |>
+ plot(main = "t()", col = rain, lwd = line_width)
+
+gpoly |> shear(fx = 0.5) |>
+ plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
+par(p)
@@ -617,20 +617,20 @@ 13.4 Affine transforms# create affine2d
-aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
+# create affine2d
+aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set.
-
-# append several simple transforms
-aff <- aff |>
- spatShift(dx = 1000) |>
- spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- flip(direction = "vertical") |>
- t() |>
- shear(fx = 0.5)
-force(aff)
+
+# append several simple transforms
+aff <- aff |>
+ spatShift(dx = 1000) |>
+ spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ flip(direction = "vertical") |>
+ t() |>
+ shear(fx = 0.5)
+force(aff)
<affine2d>
anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax)
rotate : -0.785398163397448 (rad)
@@ -639,35 +639,35 @@ 13.4 Affine transformsplot(aff)
+
We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests.
-
+
13.5 Image transforms
Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting.
spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image.
-# affine transform of image info matches with polygon info
-gimg |> affine(aff) |> plot()
-gpoly |> affine(aff) |>
- plot(add = TRUE,
- border = "cyan",
- lwd = 0.3)
+# affine transform of image info matches with polygon info
+gimg |> affine(aff) |> plot()
+gpoly |> affine(aff) |>
+ plot(add = TRUE,
+ border = "cyan",
+ lwd = 0.3)
-# affine of the giotto object
-g |> affine(aff) |>
- spatInSituPlotPoints(
- show_image = TRUE,
- feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
- feats_color_code = rainbow(4),
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- point_size = 0.1,
- use_overlap = FALSE
- )
+# affine of the giotto object
+g |> affine(aff) |>
+ spatInSituPlotPoints(
+ show_image = TRUE,
+ feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
+ feats_color_code = rainbow(4),
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ point_size = 0.1,
+ use_overlap = FALSE
+ )
Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used.
@@ -687,256 +687,256 @@ 13.6.1 Example dataset: Xenium Br
– cooresponding centroid locations
The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance.
-library(Giotto)
-instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
- save_plot = TRUE,
- show_plot = TRUE)
-options(timeout = 999999)
-download_dir <-file.path(getwd(),'/data/03_session2/')
-destfile <- file.path(download_dir,'Multimodal_registration.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
-Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
-Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
+library(Giotto)
+instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
+ save_plot = TRUE,
+ show_plot = TRUE)
+options(timeout = 999999)
+download_dir <-file.path(getwd(),'/data/03_session2/')
+destfile <- file.path(download_dir,'Multimodal_registration.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
+Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
+Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
13.6.2 Target Coordinate system
Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system.
-xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
-xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
-xen_cell_pl
+xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
+xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+xen_cell_pl
13.6.3 Visium to register
Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration
-G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
- gene_column_index = 2,
- png_name = 'tissue_hires_image.png',
- instructions = NULL)
-# In the meantime, calculate statistics for easier plot showing
-G_visium <- normalizeGiotto(G_visium)
-G_visium <- addStatistics(G_visium)
-V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
-V_origin
+G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
+ gene_column_index = 2,
+ png_name = 'tissue_hires_image.png',
+ instructions = NULL)
+# In the meantime, calculate statistics for easier plot showing
+G_visium <- normalizeGiotto(G_visium)
+G_visium <- addStatistics(G_visium)
+V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
+V_origin
The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform.
-# create affine2d
-aff <- affine(diag(c(1,1)))
-aff <- aff |>
- spin(90) |>
- flip(direction = "horizontal")
-force(aff)
-
-# Apply the transform
-V_tansformed <- affine(G_visium,aff)
-spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
-spatplot_to_register
+# create affine2d
+aff <- affine(diag(c(1,1)))
+aff <- aff |>
+ spin(90) |>
+ flip(direction = "horizontal")
+force(aff)
+
+# Apply the transform
+V_tansformed <- affine(G_visium,aff)
+spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
+spatplot_to_register
Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images.
Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app.
-
+
Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result.
-landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
-affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
-
-V_final <- affine(G_visium,affine_mtx %*% aff@affine)
-
-spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
-spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
+affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
+
+V_final <- affine(G_visium,affine_mtx %*% aff@affine)
+
+spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
+spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.3.1 Create Pseudo Visium dataset for comparison
Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance.
-V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
-spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
-spatPlot2D(V_final,
- show_network = T,
- network_color = 'blue',
- point_size = 1)
-center_to_center <- min(spat_network$distance)
-radius <- center_to_center*55/200
+V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
+spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
+spatPlot2D(V_final,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1)
+center_to_center <- min(spat_network$distance)
+radius <- center_to_center*55/200
Now we get the Pseudo Visium polygons
-Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
-stamp_dt <- circleVertices(radius = radius, npoints = 100)
-pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
-pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
-plot(pseudo_visium_poly)
+Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
+stamp_dt <- circleVertices(radius = radius, npoints = 100)
+pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
+pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
+plot(pseudo_visium_poly)
Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points
-xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
-gpoints <- createGiottoPoints(xen_transcripts)
-Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
- gpolygons = list('visium' = pseudo_visium_poly))
+xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
+gpoints <- createGiottoPoints(xen_transcripts)
+Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
+ gpolygons = list('visium' = pseudo_visium_poly))
Get gene expression information by overlapping polygon to points
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'visium')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "visium",
- feat_info = "rna",
- aggr_function = "sum")
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'visium')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "visium",
+ feat_info = "rna",
+ aggr_function = "sum")
Manipulate the expression for plotting
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'visium',
- name = 'plot')
-
-spatFeatPlot2D(Xen_obj,
- point_size = 3.5,
- expression_values = 'plot',
- show_image = F,
- feats = 'ERBB2')
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ name = 'plot')
+
+spatFeatPlot2D(Xen_obj,
+ point_size = 3.5,
+ expression_values = 'plot',
+ show_image = F,
+ feats = 'ERBB2')
Subset the registered Visium and plot same gene
-#get the extent of giotto points, xmin, xmax, ymin, ymax
-subset_extent <- ext(gpoints@spatVector)
-sub_visium <- subsetGiottoLocs(V_final,
- x_min = subset_extent[1],
- x_max = subset_extent[2],
- y_min = subset_extent[3],
- y_max = subset_extent[4])
-spatFeatPlot2D(sub_visium,
- point_size = 2,
- expression_values = 'scaled',
- show_image = F,
- feats = 'ERBB2')
+#get the extent of giotto points, xmin, xmax, ymin, ymax
+subset_extent <- ext(gpoints@spatVector)
+sub_visium <- subsetGiottoLocs(V_final,
+ x_min = subset_extent[1],
+ x_max = subset_extent[2],
+ y_min = subset_extent[3],
+ y_max = subset_extent[4])
+spatFeatPlot2D(sub_visium,
+ point_size = 2,
+ expression_values = 'scaled',
+ show_image = F,
+ feats = 'ERBB2')
13.6.4 Register post-Xenium H&E and IF image
For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics.
-HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
- imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
- micron = 0.2125)
-plot(HE_xen)
+HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
+ imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
+ micron = 0.2125)
+plot(HE_xen)
The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use.
-HE_xen <- HE_xen |> flip(direction = "vertical")
-gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
-plot(gimg_rast)
-#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
+HE_xen <- HE_xen |> flip(direction = "vertical")
+gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
+plot(gimg_rast)
+#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots
-gg <- ggplot2::ggplot()
-pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
-pl + ggplot2::geom_smooth() +
- ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+gg <- ggplot2::ggplot()
+pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
+pl + ggplot2::geom_smooth() +
+ ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.4.1 Add registered image information and compare RNA vs protein expression
With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF
-CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
-HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
-Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
- largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
+CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
+HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
+Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
+ largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
Get the cell polygons, as Xenium and IF are both subcellular resolution
-cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
-colnames(cellpoly_dt) <- c('poly_ID','x','y')
-cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
-Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
+cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
+colnames(cellpoly_dt) <- c('poly_ID','x','y')
+cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
+Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
Compute the gene expression matrix by overlay the cell polygons and giotto points.
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'cell')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "cell",
- feat_info = "rna",
- aggr_function = "sum")
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'rna',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'ERBB2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'cell')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "cell",
+ feat_info = "rna",
+ aggr_function = "sum")
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'rna',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'ERBB2',
+ point_size = 0.05)
Now we overlay the HER2 expression from the raster image with the cell polygons.
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'cell',
- image_names = c('HER2','CD20'))
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'HER2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'cell',
+ image_names = c('HER2','CD20'))
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'HER2',
+ point_size = 0.05)
We can also overlay the protein expression to Visium spots
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'visium',
- image_names = c('HER2','CD20'))
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "visium",
- feat_info = "protein",
- aggr_function = "sum")
-
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'visium',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'visium',
- feats = 'HER2',
- point_size = 2)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'visium',
+ image_names = c('HER2','CD20'))
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "visium",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'visium',
+ feats = 'HER2',
+ point_size = 2)
@@ -944,29 +944,29 @@ 13.6.4.1 Add registered image inf
13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation
Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially.
Here, we provide an example of two compressed images to show the automatic alignment pipeline.
-
+
-IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
-terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
+IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
+terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform.
-estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
- flip_horizontal = T,
- use_single_channel = T,
- single_channel_number = 3),
- y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
- invert = T),
- plot_match = T,
- max_ratio = 0.5,estimate_fun = 'Projective')
+estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
+ flip_horizontal = T,
+ use_single_channel = T,
+ single_channel_number = 3),
+ y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
+ invert = T),
+ plot_match = T,
+ max_ratio = 0.5,estimate_fun = 'Projective')
Use the estimation, we can quickly visualize the transformation
-mtx <- as.matrix(estimation$params)
-transformed <- affine(IF, mtx)
-To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
-plot(HE)
-plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
-SIFT_registration_overlay
+mtx <- as.matrix(estimation$params)
+transformed <- affine(IF, mtx)
+To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
+plot(HE)
+plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
+SIFT_registration_overlay
diff --git a/spatial-proteomics-multiplexed-immunofluorescence.html b/spatial-proteomics-multiplexed-immunofluorescence.html
index 684d4e8..f550962 100644
--- a/spatial-proteomics-multiplexed-immunofluorescence.html
+++ b/spatial-proteomics-multiplexed-immunofluorescence.html
@@ -518,33 +518,33 @@ 11 Spatial proteomics: Multiplexe
Junxiang Xu
August 6th 2024
Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so.
-#Install the Giotto Environment with Cellpose, note that we only need to do it once
-reticulate::conda_create(envname = "giotto_cellpose",
- python_version = 3.8)
-#.re.restartR()
-reticulate::use_condaenv('giotto_cellpose')
-reticulate::py_install(
- pip = TRUE,
- envname = 'giotto_cellpose',
- packages = c(
- "pandas",
- "networkx",
- "python-igraph",
- "leidenalg",
- "scikit-learn",
- "cellpose",
- "smfishhmrf",
- 'tifffile',
- 'scikit-image'
- )
-)
-#.rs.restartR()
+#Install the Giotto Environment with Cellpose, note that we only need to do it once
+reticulate::conda_create(envname = "giotto_cellpose",
+ python_version = 3.8)
+#.re.restartR()
+reticulate::use_condaenv('giotto_cellpose')
+reticulate::py_install(
+ pip = TRUE,
+ envname = 'giotto_cellpose',
+ packages = c(
+ "pandas",
+ "networkx",
+ "python-igraph",
+ "leidenalg",
+ "scikit-learn",
+ "cellpose",
+ "smfishhmrf",
+ 'tifffile',
+ 'scikit-image'
+ )
+)
+#.rs.restartR()
Now, activate the Giotto python environment.
-#.rs.restartR()
-# Activate the Giotto python environment of your choice
-GiottoClass::set_giotto_python_path('giotto_cellpose')
-# Check if cellpose was successfully installed
-GiottoUtils::package_check('cellpose',repository = 'pip')
+#.rs.restartR()
+# Activate the Giotto python environment of your choice
+GiottoClass::set_giotto_python_path('giotto_cellpose')
+# Check if cellpose was successfully installed
+GiottoUtils::package_check('cellpose',repository = 'pip')
11.1 Spatial Proteomics Technologies
This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix.
@@ -557,58 +557,58 @@
11.2 Raw data type coming out of
11.2.1 Use ome.tiff as an example output data to begin with
OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore.
-library(Giotto)
-instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
- save_plot = TRUE,
- show_plot = TRUE,
- python_path = 'giotto_cellpose')
-options(timeout = 9999999)
-download_dir =file.path(getwd(),'/data/02_session4/')
-destfile = file.path(download_dir,'Lunaphore.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
-list.files(paste0(download_dir,'/Lunaphore'))
+library(Giotto)
+instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ python_path = 'giotto_cellpose')
+options(timeout = 9999999)
+download_dir =file.path(getwd(),'/data/02_session4/')
+destfile = file.path(download_dir,'Lunaphore.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
+list.files(paste0(download_dir,'/Lunaphore'))
We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML.
-img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
-img_meta
+img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
+img_meta
However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it.
-##Get channel information
-Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
-node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
-channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
-channel_df
+##Get channel information
+Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
+node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
+channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
+channel_df
11.2.2 Use single channel images as an example output data to begin with
Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like.
-#Create multichannel raster and extract each single channels
-Luna_terra = terra::rast(Luna)
-names(Luna_terra) <- channel_df$Name
-gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
-plot(gimg_DAPI)
+#Create multichannel raster and extract each single channels
+Luna_terra = terra::rast(Luna)
+names(Luna_terra) <- channel_df$Name
+gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
+plot(gimg_DAPI)
Extract and save the raster image for future use.
-single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
-if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
-for (i in 1:nrow(channel_df)){
- single_channel <- terra::subset(Luna_terra, i)
- terra::writeRaster(single_channel,
- filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
- overwrite=T)
-}
+single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
+if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
+for (i in 1:nrow(channel_df)){
+ single_channel <- terra::subset(Luna_terra, i)
+ terra::writeRaster(single_channel,
+ filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
+ overwrite=T)
+}
Create a list of GiottoLargeImages using single channel rasters.
-file_names = list.files(single_channel_dir,full.names = TRUE)
-image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
-gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
- names = image_names,
- negative_y = F,
- flip_vertical = T)
-names(gimg_list) <- image_names
-plot(gimg_list[['Vimentin']])
+file_names = list.files(single_channel_dir,full.names = TRUE)
+image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
+gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
+ names = image_names,
+ negative_y = F,
+ flip_vertical = T)
+names(gimg_list) <- image_names
+plot(gimg_list[['Vimentin']])
@@ -618,34 +618,34 @@ 11.3 Cell Segmentation file to ge
11.3.1 Using segmentation output file from DeepCell(mesmer) as an example.
We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function.
-gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
-plot(gpoly_mesmer)
+gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
+plot(gpoly_mesmer)
We can also zoom in to check how does the segmentation look.
-zoom <- c(2000,2500,2000,2500)
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+zoom <- c(2000,2500,2000,2500)
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
11.3.2 Using Giotto wrapper of Cellpose to perform segmentation
-gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
-writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
-#Create a giotto image to evaluate segmentation
-gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
+gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
+writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
+#Create a giotto image to evaluate segmentation
+gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website
-doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
- mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- channel_1 = 0,
- channel_2 = 0,
- model_name = 'cyto3',
- batch_size=12)
-cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- shift_horizontal_step = F,
- shift_vertical_step = F,
- flip_vertical = T)
-plot(gimg_for_cellpose)
-plot(cpoly, add = T, border = 'red')
+doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
+ mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ channel_1 = 0,
+ channel_2 = 0,
+ model_name = 'cyto3',
+ batch_size=12)
+cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ shift_horizontal_step = F,
+ shift_vertical_step = F,
+ flip_vertical = T)
+plot(gimg_for_cellpose)
+plot(cpoly, add = T, border = 'red')
@@ -654,133 +654,133 @@ 11.4 Create a Giotto Object using
You will need to have
- list of giotto images
- giottoPolygon created from segmentation
-Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
- images = gimg_list,
- instructions = instrs)
-Lunaphore_giotto
+Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
+ images = gimg_list,
+ instructions = instrs)
+Lunaphore_giotto
11.4.1 Overlap to matrix
calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with
-Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
- spatial_info = 'cell',
- image_names = names(gimg_list))
-
-Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
- type ='intensity',
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-showGiottoExpression(Lunaphore_giotto)
+Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
+ spatial_info = 'cell',
+ image_names = names(gimg_list))
+
+Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
+ type ='intensity',
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+showGiottoExpression(Lunaphore_giotto)
11.4.2 Manipulate Expression information
For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile.
Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object
-expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
-filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
-Lunaphore_giotto <- setExpression(Lunaphore_giotto,
- feat_type = 'protein',
- x = createExprObj(filtered_expr_mtx),
- name = 'raw')
-showGiottoExpression(Lunaphore_giotto)
+expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
+filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
+Lunaphore_giotto <- setExpression(Lunaphore_giotto,
+ feat_type = 'protein',
+ x = createExprObj(filtered_expr_mtx),
+ name = 'raw')
+showGiottoExpression(Lunaphore_giotto)
11.4.3 Rescale polygons
rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis
-Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
- poly_info = 'cell',
- name = 'smallcell',
- fx = 0.7, fy = 0.7, calculate_centroids = T)
-smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
-plot(smallpoly,add = TRUE, border = "red",ext = zoom)
+Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
+ poly_info = 'cell',
+ name = 'smallcell',
+ fx = 0.7, fy = 0.7, calculate_centroids = T)
+smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+plot(smallpoly,add = TRUE, border = "red",ext = zoom)
11.4.4 Perform clustering and differential expression
The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering
-Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
-Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
-
-Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
- scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
- set_seed = TRUE)
-screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
+Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
+Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
+
+Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
+ scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
+ set_seed = TRUE)
+screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check.
-Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
- dimensions_to_use = 1:5,
- set_seed = TRUE)
-Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
-Lunaphore_giotto = doHclust(Lunaphore_giotto,
- k = 8,
- dim_reduction_to_use = 'cells',
- spat_unit = 'cell',
- feat_type = 'protein')
-
-spatInSituPlotPoints(gobject = Lunaphore_giotto,
- spat_unit = "cell",
- polygon_feat_type = "cell",
- show_polygon = T,
- feat_type = "protein",
- feats = NULL,
- polygon_fill = 'hclust',
- polygon_fill_as_factor = TRUE,
- polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
- polygon_color = 'black',
- background_color = "white")
+Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
+ dimensions_to_use = 1:5,
+ set_seed = TRUE)
+Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
+Lunaphore_giotto = doHclust(Lunaphore_giotto,
+ k = 8,
+ dim_reduction_to_use = 'cells',
+ spat_unit = 'cell',
+ feat_type = 'protein')
+
+spatInSituPlotPoints(gobject = Lunaphore_giotto,
+ spat_unit = "cell",
+ polygon_feat_type = "cell",
+ show_polygon = T,
+ feat_type = "protein",
+ feats = NULL,
+ polygon_fill = 'hclust',
+ polygon_fill_as_factor = TRUE,
+ polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
+ polygon_color = 'black',
+ background_color = "white")
Then we can check the heatmap of protein expression and determine the first round of cluster annotation
-cluster_column = 'hclust'
-plotMetaDataHeatmap(Lunaphore_giotto,
- spat_unit = 'cell',feat_type = 'protein',
- expression_values = "raw",
- metadata_cols = c(cluster_column),
- selected_feats = names(gimg_list),
- y_text_size = 8,
- show_values = 'zscores_rescaled')
+cluster_column = 'hclust'
+plotMetaDataHeatmap(Lunaphore_giotto,
+ spat_unit = 'cell',feat_type = 'protein',
+ expression_values = "raw",
+ metadata_cols = c(cluster_column),
+ selected_feats = names(gimg_list),
+ y_text_size = 8,
+ show_values = 'zscores_rescaled')
11.4.5 Give the cluster an annotation based on expression values
-
+
11.4.6 spatial network
This is to create a cellular neighborhood based on nearest neighbor of physical distance
-Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
-
-spatPlot2D(Lunaphore_giotto,
- show_network = T,
- network_color = 'blue',
- point_size = 1.5,
- cell_color = 'hclust')
+Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
+
+spatPlot2D(Lunaphore_giotto,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1.5,
+ cell_color = 'hclust')
11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions
This is using cellProximityEnrichment() to statistically identify cell type interactions
-cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
- cluster_column = 'cell_types',
- spatial_network_name = 'Delaunay_network',
- adjust_method = 'fdr',
- number_of_simulations = 2000)
-## barplot
-cellProximityBarplot(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- min_orig_ints = 5, min_sim_ints = 5)
+cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
+ cluster_column = 'cell_types',
+ spatial_network_name = 'Delaunay_network',
+ adjust_method = 'fdr',
+ number_of_simulations = 2000)
+## barplot
+cellProximityBarplot(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ min_orig_ints = 5, min_sim_ints = 5)
-## network
-cellProximityNetwork(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- remove_self_edges = T,
- only_show_enrichment_edges = F)
+## network
+cellProximityNetwork(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ remove_self_edges = T,
+ only_show_enrichment_edges = F)
diff --git a/visium-hd.html b/visium-hd.html
index 23aa210..f24cdd4 100644
--- a/visium-hd.html
+++ b/visium-hd.html
@@ -525,7 +525,7 @@ 9.1 Objective9.2 Background
9.2.1 Visium HD Technology
-
+
Figure 9.1: Overview of Visium HD. Source: 10X Genomics
@@ -536,7 +536,7 @@
9.2.1 Visium HD Technology
9.2.2 Colorectal Cancer Sample
-
+
Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics
@@ -550,7 +550,7 @@
9.2.2 Colorectal Cancer Sample9.3 Data Ingestion
9.3.1 Visium HD output data format
-
+
Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline.
@@ -560,19 +560,19 @@
9.3.1 Visium HD output data forma
9.4 Tiling and aggregation
The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells.
-
+
9.5 Scalability and projection functions
diff --git a/visium-part-i.html b/visium-part-i.html
index 5aa1658..c4746ee 100644
--- a/visium-part-i.html
+++ b/visium-part-i.html
@@ -520,7 +520,7 @@ 7 Visium Part I
7.1 The Visium technology
Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression.
-
+
Figure 7.1: Visum workflow. Source: 10X Genomics
@@ -536,73 +536,73 @@
7.2 Introduction to the spatial d
7.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/01_session5/")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
+dir.create("data/01_session5/")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”.
-
+
7.4 Create the Giotto object
createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/01_session5/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-## Provide the path to the visium folder
-data_path <- "data/01_session5/"
-
-## Create object directly from the visium folder
-visium_brain <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/01_session5/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+## Provide the path to the visium folder
+data_path <- "data/01_session5/"
+
+## Create object directly from the visium folder
+visium_brain <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
7.5 Subset on spots that were covered by tissue
Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area.
-spatPlot2D(
- gobject = visium_brain,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE)
-
+spatPlot2D(
+ gobject = visium_brain,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE)
+
Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area.
-
+
7.6 Quality control
@@ -610,43 +610,43 @@ 7.6 Quality controlvisium_brain_statistics <- addStatistics(gobject = visium_brain,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = visium_brain_statistics,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain_statistics <- addStatistics(gobject = visium_brain,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = visium_brain_statistics,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.3: Spatial distribution of features per spot.
filterDistributions() creates a histogram to show the distribution of features per spot across the sample.
-
-
+
+
Figure 7.4: Distribution of features per spot.
When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample.
-
-
+
+
Figure 7.5: Distribution of cells with different features per spot.
filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data:
-filterCombinations(gobject = visium_brain_statistics,
- expression_thresholds = c(1, 2, 3),
- feat_det_in_min_cells = c(50, 100, 200),
- min_det_feats_per_cell = c(500, 1000, 1500))
-
+filterCombinations(gobject = visium_brain_statistics,
+ expression_thresholds = c(1, 2, 3),
+ feat_det_in_min_cells = c(50, 100, 200),
+ min_det_feats_per_cell = c(500, 1000, 1500))
+
Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations.
@@ -656,34 +656,34 @@
7.6 Quality control
7.7 Filtering
Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample.
-visium_brain <- filterGiotto(
- gobject = visium_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE
-)
-Feature type: rna
-Number of cells removed: 4 out of 2702
-Number of feats removed: 7311 out of 22125
+
+
7.8 Normalization
Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one.
-visium_brain <- normalizeGiotto(
- gobject = visium_brain,
- scalefactor = 6000,
- verbose = TRUE
-)
+visium_brain <- normalizeGiotto(
+ gobject = visium_brain,
+ scalefactor = 6000,
+ verbose = TRUE
+)
Calculate the normalized number of features per spot and save the statistics in the metadata table.
-visium_brain <- addStatistics(gobject = visium_brain)
-
-## visualize
-spatPlot2D(gobject = visium_brain,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain <- addStatistics(gobject = visium_brain)
+
+## visualize
+spatPlot2D(gobject = visium_brain,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.7: Spatial distribution of the number of features per spot.
@@ -698,11 +698,11 @@
7.9.1 Highly Variable Features:
loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_loess",
- save_plot = TRUE,
- default_save_name = "HVFplot_loess")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_loess",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_loess")
+
Figure 7.8: Covariance of HVFs using the loess method.
@@ -711,11 +711,11 @@
7.9.1 Highly Variable Features:
pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")
+
Figure 7.9: Variance of HVFs using the pearson residuals method.
@@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
Figure 17.5: Giotto object before adding subcellular polygons. @@ -697,11 +697,11 @@
17.4.3 Showing Giotto object prio
17.4.4 Adding polygons to giotto object
After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject.
-v_brain <- addGiottoPolygons(v_brain,
- gpolygons = list('stardist_cell' = stardist_cell_gpoly))
-
-print(v_brain)
-
+v_brain <- addGiottoPolygons(v_brain,
+ gpolygons = list('stardist_cell' = stardist_cell_gpoly))
+
+print(v_brain)
+
Figure 17.6: Giotto object after to adding subcellular polygons.
@@ -711,9 +711,9 @@
17.4.4 Adding polygons to giotto
17.4.5 Check polygon information
We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots.
-
-
+
+
Figure 17.7: Polygon information for stardist_cell.
@@ -727,23 +727,23 @@
17.5 Performing kriging17.5.1 Interpolating features
Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be
analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required.
-future::plan(future::multisession()) # comment out for single threading
-subcell_v_brain <- interpolateFeature(v_brain,
- spat_unit = 'cell',
- feat_type = 'rna',
- ext = ext(v_brain),
- feats = ext_spatial_features,
- overwrite = T)
-
-print(subcell_v_brain)
-
+future::plan(future::multisession()) # comment out for single threading
+subcell_v_brain <- interpolateFeature(v_brain,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ ext = ext(v_brain),
+ feats = ext_spatial_features,
+ overwrite = T)
+
+print(subcell_v_brain)
+
Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right).
For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1.
-
+
Figure 17.9: Raster of gene expression interpolation for Pantr1
@@ -754,18 +754,18 @@
17.5.1 Interpolating features17.5.2 Expression overlap
The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier.
This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run.
-subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain,
- name_overlap = "rna",
- spatial_info = "stardist_cell",
- image_names = ext_spatial_features)
-
-subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain,
- poly_info = "stardist_cell",
- feat_info = "rna",
- aggr_function = "sum",
- type='intensity')
+subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain,
+ name_overlap = "rna",
+ spatial_info = "stardist_cell",
+ image_names = ext_spatial_features)
+
+subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain,
+ poly_info = "stardist_cell",
+ feat_info = "rna",
+ aggr_function = "sum",
+ type='intensity')
After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified.
-
+
Figure 17.10: Gene expression for cells based on interpolation.
@@ -776,70 +776,70 @@
17.5.2 Expression overlap
17.6 Reading in larger dataset
For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here.
-
+
17.7 Analyzing interpolated features
17.7.1 Filter and normalization
Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis.
-subcell_v_brain <- filterGiotto(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- expression_values = "raw",
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-
-subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- scalefactor = 6000,
- verbose = T)
+subcell_v_brain <- filterGiotto(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ expression_values = "raw",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+
+subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ scalefactor = 6000,
+ verbose = T)
17.7.2 Visualizing gene expression from interpolated expression
Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data.
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Sparc',
- point_size = 2,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_sparc_vis'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'stardist_cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Sparc',
- point_size = 0.6,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_sparc'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'cell',
- gradient_style = 'sequential',
- feats = 'Pantr1',
- cell_color_gradient = 'Geyser',
- point_size = 2,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_pantr1_vis'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'stardist_cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Pantr1',
- point_size = 0.6,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_pantr1'))
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Sparc',
+ point_size = 2,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_sparc_vis'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Sparc',
+ point_size = 0.6,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_sparc'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'cell',
+ gradient_style = 'sequential',
+ feats = 'Pantr1',
+ cell_color_gradient = 'Geyser',
+ point_size = 2,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_pantr1_vis'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Pantr1',
+ point_size = 0.6,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_pantr1'))
Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer.
-
+
Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom).
@@ -848,37 +848,37 @@
17.7.2 Visualizing gene expressio
17.7.3 Run PCA
-
+
17.7.4 Clustering
-# UMAP
-subcell_v_brain <- runUMAP(subcell_v_brain,
- spat_unit = "stardist_cell",
- dimensions_to_use = 1:15,
- n_neighbors = 1000,
- min_dist = 0.001,
- spread = 1)
-
-# NN Network
-subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- dimensions_to_use = 1:10,
- feats_to_use = hv_feats,
- expression_values = 'normalized',
- k = 70)
-
-subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- resolution = 0.15,
- n_iterations = 100,
- partition_type = 'RBConfigurationVertexPartition')
-
-plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus')
-
+# UMAP
+subcell_v_brain <- runUMAP(subcell_v_brain,
+ spat_unit = "stardist_cell",
+ dimensions_to_use = 1:15,
+ n_neighbors = 1000,
+ min_dist = 0.001,
+ spread = 1)
+
+# NN Network
+subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ dimensions_to_use = 1:10,
+ feats_to_use = hv_feats,
+ expression_values = 'normalized',
+ k = 70)
+
+subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ resolution = 0.15,
+ n_iterations = 100,
+ partition_type = 'RBConfigurationVertexPartition')
+
+plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus')
+
Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering.
@@ -888,27 +888,27 @@
17.7.4 Clustering
17.7.5 Visualizing clustering
Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster.
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.5,
- point_shape = 'no_border',
- background_color = 'black',
- save_plot = F,
- show_legend = TRUE)
-
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat'))
-
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.5,
+ point_shape = 'no_border',
+ background_color = 'black',
+ save_plot = F,
+ show_legend = TRUE)
+
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat'))
+
Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
@@ -918,37 +918,37 @@
17.7.5 Visualizing clustering
17.7.6 Cropping objects
We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below.
-subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
- spat_unit = ":all:",
- x_min = 4000,
- x_max = 7000,
- y_min = -6500,
- y_max = -3500,
- z_max = NULL,
- z_min = NULL)
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 2,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_vis_spat_crop'))
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat_crop'))
-
+subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
+ spat_unit = ":all:",
+ x_min = 4000,
+ x_max = 7000,
+ y_min = -6500,
+ y_max = -3500,
+ z_max = NULL,
+ z_min = NULL)
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 2,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_vis_spat_crop'))
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat_crop'))
+
Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
diff --git a/interoperability-with-isolated-tools.html b/interoperability-with-isolated-tools.html
index ceb2d0a..8cca7e6 100644
--- a/interoperability-with-isolated-tools.html
+++ b/interoperability-with-isolated-tools.html
@@ -526,1521 +526,1521 @@
16.1.1 Dataset downloaddata_path <- "data"
-dir.create(data_path)
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
- destfile = file.path(data_path,"counts.h5ad"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
- destfile = file.path(data_path,"cell_labels.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
+data_path <- "data"
+dir.create(data_path)
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
+ destfile = file.path(data_path,"counts.h5ad"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
+ destfile = file.path(data_path,"cell_labels.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
16.1.2 Create the Giotto object
-library(Giotto)
-library(reticulate)
-
-## Set instructions
-results_folder <- "results"
-dir.create(results_folder)
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-
-## load data
-
-### meta data
-meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-colnames(meta_df)[[1]] <- "cell_ID"
-slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
-slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
-
-### counts matrix
-scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
-ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
-reticulate::source_python(ad2g_path)
-adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
-X <- extract_expression(adata)
-cID <- extract_cell_IDs(adata)
-fID <- extract_feat_IDs(adata)
-rownames(X) <- fID
-colnames(X) <- cID
-slice1_filter <- cID %in% slice1_meta_df$cell_ID
-slice2_filter <- cID %in% slice2_meta_df$cell_ID
-slice1_exp <- X[,slice1_filter]
-slice2_exp <- X[,slice2_filter]
-
-### cell segmentation to cell location
-segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_df <- rbind(segments_1_df, segments_2_df)
-loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
-loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
-loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
-centr.x <- apply(loc.use[,loc.x],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-centr.y <- apply(loc.use[,loc.y],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
-slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
-slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
-
-## giotto obj
-giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
- spatial_locs = slice1_spatial_locs_df,
- cell_metadata = slice1_meta_df)
-giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
- spatial_locs = slice2_spatial_locs_df,
- cell_metadata = slice2_meta_df)
-giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
- gobject_names = c('mouse2_slice229',
- 'mouse2_slice300'), # name for each samples
- join_method = 'z_stack')
-# We skip the processing process here to save time and use the given cell type
-# annotation directly
-
-ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
- cell_type = "subclass",
- output_path = data_path,
- spat_unit = "cell",
- feat_type = "rna",
- verbose = TRUE)
-
-# Cell_ID Sample x y Cell_Type
-# <char> <char> <dbl> <dbl> <char>
-# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
-# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
-# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
-# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
-# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
-# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
+library(Giotto)
+library(reticulate)
+
+## Set instructions
+results_folder <- "results"
+dir.create(results_folder)
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+
+## load data
+
+### meta data
+meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+colnames(meta_df)[[1]] <- "cell_ID"
+slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
+slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
+
+### counts matrix
+scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
+ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
+reticulate::source_python(ad2g_path)
+adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
+X <- extract_expression(adata)
+cID <- extract_cell_IDs(adata)
+fID <- extract_feat_IDs(adata)
+rownames(X) <- fID
+colnames(X) <- cID
+slice1_filter <- cID %in% slice1_meta_df$cell_ID
+slice2_filter <- cID %in% slice2_meta_df$cell_ID
+slice1_exp <- X[,slice1_filter]
+slice2_exp <- X[,slice2_filter]
+
+### cell segmentation to cell location
+segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_df <- rbind(segments_1_df, segments_2_df)
+loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
+loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
+loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
+centr.x <- apply(loc.use[,loc.x],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+centr.y <- apply(loc.use[,loc.y],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
+slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
+slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
+
+## giotto obj
+giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
+ spatial_locs = slice1_spatial_locs_df,
+ cell_metadata = slice1_meta_df)
+giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
+ spatial_locs = slice2_spatial_locs_df,
+ cell_metadata = slice2_meta_df)
+giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
+ gobject_names = c('mouse2_slice229',
+ 'mouse2_slice300'), # name for each samples
+ join_method = 'z_stack')
+# We skip the processing process here to save time and use the given cell type
+# annotation directly
+
+ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
+ cell_type = "subclass",
+ output_path = data_path,
+ spat_unit = "cell",
+ feat_type = "rna",
+ verbose = TRUE)
+
+# Cell_ID Sample x y Cell_Type
+# <char> <char> <dbl> <dbl> <char>
+# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
+# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
+# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
+# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
+# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
+# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
Spatial cell type distributions
-spatPlot2D(giotto_obj,
- group_by = "slice_id",
- cell_color = "subclass",
- point_size = 1,
- point_border_stroke = NA,
- legend_text = 6)
+spatPlot2D(giotto_obj,
+ group_by = "slice_id",
+ cell_color = "subclass",
+ point_size = 1,
+ point_border_stroke = NA,
+ legend_text = 6)
16.1.2.1 Installation ONTraC
You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node.
It will run for less than 10 minutes on this example dataset.
For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time.
-source ~/.bash_profile
-conda create -y -n ONTraC python=3.11
-conda activate ONTraC
-pip install git+https://github.com/gyuanlab/ONTraC.git@V2
-# Retrieving notices: ...working... done
-# Channels:
-# - conda-forge
-# - bioconda
-# - r
-# - pytorch
-# - defaults
-# Platform: osx-arm64
-# Collecting package metadata (repodata.json): ...working... done
-# Solving environment: ...working... done
-#
-# ## Package Plan ##
-#
-# environment location: /Users/wwang/miniconda3/envs/ONTraC
-#
-# added / updated specs:
-# - python=3.11
-#
-#
-# The following packages will be downloaded:
-#
-# package | build
-# ---------------------------|-----------------
-# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
-# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
-# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
-# ------------------------------------------------------------
-# Total: 4.3 MB
-#
-# The following NEW packages will be INSTALLED:
-#
-# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
-# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
-# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
-# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
-# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
-# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
-# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
-# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
-# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
-# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
-# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
-# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
-# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
-# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
-# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
-# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
-#
-#
-#
-# Downloading and Extracting Packages: ...working... done
-# Preparing transaction: ...working... done
-# Verifying transaction: ...working... done
-# Executing transaction: ...working... done
-# #
-# # To activate this environment, use
-# #
-# # $ conda activate ONTraC
-# #
-# # To deactivate an active environment, use
-# #
-# # $ conda deactivate
-#
-# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
-# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
-# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
-# Running command git checkout -b V2 --track origin/V2
-# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
-# Installing build dependencies: started
-# Switched to a new branch 'V2'
-# branch 'V2' set up to track 'origin/V2'.
-# Installing build dependencies: finished with status 'done'
-# Getting requirements to build wheel: started
-# Getting requirements to build wheel: finished with status 'done'
-# Preparing metadata (pyproject.toml): started
-# Preparing metadata (pyproject.toml): finished with status 'done'
-# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
-# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
-# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
-# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
-# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
-# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
-# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
-# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
-# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
-# Collecting session-info (from ONTraC==2.0a9.dev7)
-# Using cached session_info-1.0.0-py3-none-any.whl
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
-# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
-# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
-# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
-# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
-# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
-# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
-# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
-# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
-# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
-# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
-# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
-# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
-# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
-# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
-# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
-# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
-# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
-# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
-# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
-# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
-# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
-# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
-# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
-# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
-# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
-# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
-# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
-# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
-# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
-# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
-# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
-# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
-# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
-# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
-# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
-# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
-# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
-# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
-# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
-# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
-# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
-# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
-# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
-# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
-# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
-# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
-# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
-# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
-# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
-# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
-# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
-# Using cached idna-3.7-py3-none-any.whl (66 kB)
-# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
-# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
-# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
-# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
-# Building wheels for collected packages: ONTraC
-# Building wheel for ONTraC (pyproject.toml): started
-# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
-# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
-# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
-# Successfully built ONTraC
-# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
-# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
+source ~/.bash_profile
+conda create -y -n ONTraC python=3.11
+conda activate ONTraC
+pip install git+https://github.com/gyuanlab/ONTraC.git@V2
+# Retrieving notices: ...working... done
+# Channels:
+# - conda-forge
+# - bioconda
+# - r
+# - pytorch
+# - defaults
+# Platform: osx-arm64
+# Collecting package metadata (repodata.json): ...working... done
+# Solving environment: ...working... done
+#
+# ## Package Plan ##
+#
+# environment location: /Users/wwang/miniconda3/envs/ONTraC
+#
+# added / updated specs:
+# - python=3.11
+#
+#
+# The following packages will be downloaded:
+#
+# package | build
+# ---------------------------|-----------------
+# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
+# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
+# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
+# ------------------------------------------------------------
+# Total: 4.3 MB
+#
+# The following NEW packages will be INSTALLED:
+#
+# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
+# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
+# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
+# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
+# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
+# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
+# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
+# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
+# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
+# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
+# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
+# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
+# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
+# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
+# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
+# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
+#
+#
+#
+# Downloading and Extracting Packages: ...working... done
+# Preparing transaction: ...working... done
+# Verifying transaction: ...working... done
+# Executing transaction: ...working... done
+# #
+# # To activate this environment, use
+# #
+# # $ conda activate ONTraC
+# #
+# # To deactivate an active environment, use
+# #
+# # $ conda deactivate
+#
+# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
+# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
+# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
+# Running command git checkout -b V2 --track origin/V2
+# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
+# Installing build dependencies: started
+# Switched to a new branch 'V2'
+# branch 'V2' set up to track 'origin/V2'.
+# Installing build dependencies: finished with status 'done'
+# Getting requirements to build wheel: started
+# Getting requirements to build wheel: finished with status 'done'
+# Preparing metadata (pyproject.toml): started
+# Preparing metadata (pyproject.toml): finished with status 'done'
+# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
+# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
+# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
+# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
+# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
+# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
+# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
+# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
+# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
+# Collecting session-info (from ONTraC==2.0a9.dev7)
+# Using cached session_info-1.0.0-py3-none-any.whl
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
+# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
+# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
+# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
+# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
+# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
+# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
+# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
+# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
+# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
+# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
+# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
+# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
+# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
+# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
+# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
+# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
+# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
+# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
+# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
+# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
+# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
+# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
+# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
+# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
+# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
+# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
+# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
+# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
+# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
+# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
+# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
+# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
+# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
+# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
+# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
+# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
+# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
+# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
+# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
+# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
+# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
+# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
+# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
+# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
+# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
+# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
+# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
+# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
+# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
+# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
+# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
+# Using cached idna-3.7-py3-none-any.whl (66 kB)
+# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
+# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
+# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
+# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
+# Building wheels for collected packages: ONTraC
+# Building wheel for ONTraC (pyproject.toml): started
+# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
+# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
+# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
+# Successfully built ONTraC
+# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
+# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
16.1.2.2 Running ONTraC
-
-source ~/.bash_profile
-conda activate ONTraC_v2_py311
-ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
-# ##################################################################################
-#
-# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
-# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
-# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
-# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
-# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
-#
-# version: 2.0a11
-#
-# ##################################################################################
-# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
-# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
-# 11:33:23 --- INFO: -------- I/O options -------
-# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
-# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
-# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
-# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
-# 11:33:23 --- INFO: -------- preprocessing options -------
-# 11:33:23 --- INFO: resolution: 10.0
-# 11:33:23 --- INFO: -------- niche net constr options -------
-# 11:33:23 --- INFO: n_cpu: 4
-# 11:33:23 --- INFO: n_neighbors: 50
-# 11:33:23 --- INFO: n_local: 20
-# 11:33:23 --- INFO: embedding_adjust: False
-# 11:33:23 --- INFO: sigma: 1
-# 11:33:23 --- INFO: -------- train options -------
-# 11:33:23 --- INFO: device: cpu
-# 11:33:23 --- INFO: epochs: 1000
-# 11:33:23 --- INFO: batch_size: 0
-# 11:33:23 --- INFO: patience: 100
-# 11:33:23 --- INFO: min_delta: 0.001
-# 11:33:23 --- INFO: min_epochs: 50
-# 11:33:23 --- INFO: seed: 42
-# 11:33:23 --- INFO: lr: 0.03
-# 11:33:23 --- INFO: hidden_feats: 4
-# 11:33:23 --- INFO: k: 6
-# 11:33:23 --- INFO: modularity_loss_weight: 1.0
-# 11:33:23 --- INFO: purity_loss_weight: 100.0
-# 11:33:23 --- INFO: regularization_loss_weight: 0.1
-# 11:33:23 --- INFO: beta: 0.3
-# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
-# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
-# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
-# 11:33:23 --- INFO: ------------- Niche network construct ---------------
-# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
-# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
-# 11:33:28 --- INFO: Generating samples.yaml file.
-# 11:33:28 --- INFO: ------------ Niche network construct end ------------
-# 11:33:28 --- INFO: ------------------------ GNN ------------------------
-# 11:33:28 --- INFO: Loading dataset.
-# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
-# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
-# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
-# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
-# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
-# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
-# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
-# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
-# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
-# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
-# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
-# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
-# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
-# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
-# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
-# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
-# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
-# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
-# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
-# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
-# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
-# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
-# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
-# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
-# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
-# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
-# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
-# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
-# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
-# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
-# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
-# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
-# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
-# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
-# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
-# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
-# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
-# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
-# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
-# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
-# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
-# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
-# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
-# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
-# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
-# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
-# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
-# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
-# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
-# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
-# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
-# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
-# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
-# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
-# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
-# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
-# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
-# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
-# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
-# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
-# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
-# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
-# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
-# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
-# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
-# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
-# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
-# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
-# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
-# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
-# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
-# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
-# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
-# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
-# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
-# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
-# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
-# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
-# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
-# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
-# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
-# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
-# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
-# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
-# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
-# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
-# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
-# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
-# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
-# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
-# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
-# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
-# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
-# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
-# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
-# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
-# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
-# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
-# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
-# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
-# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
-# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
-# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
-# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
-# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
-# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
-# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
-# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
-# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
-# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
-# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
-# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
-# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
-# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
-# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
-# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
-# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
-# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
-# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
-# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
-# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
-# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
-# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
-# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
-# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
-# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
-# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
-# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
-# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
-# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
-# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
-# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
-# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
-# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
-# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
-# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
-# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
-# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
-# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
-# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
-# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
-# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
-# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
-# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
-# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
-# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
-# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
-# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
-# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
-# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
-# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
-# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
-# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
-# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
-# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
-# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
-# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
-# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
-# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
-# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
-# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
-# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
-# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
-# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
-# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
-# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
-# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
-# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
-# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
-# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
-# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
-# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
-# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
-# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
-# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
-# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
-# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
-# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
-# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
-# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
-# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
-# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
-# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
-# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
-# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
-# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
-# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
-# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
-# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
-# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
-# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
-# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
-# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
-# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
-# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
-# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
-# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
-# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
-# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
-# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
-# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
-# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
-# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
-# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
-# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
-# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
-# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
-# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
-# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
-# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
-# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
-# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
-# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
-# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
-# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
-# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
-# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
-# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
-# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
-# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
-# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
-# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
-# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
-# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
-# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
-# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
-# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
-# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
-# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
-# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
-# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
-# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
-# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
-# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
-# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
-# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
-# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
-# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
-# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
-# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
-# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
-# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
-# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
-# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
-# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
-# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
-# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
-# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
-# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
-# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
-# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
-# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
-# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
-# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
-# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
-# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
-# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
-# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
-# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
-# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
-# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
-# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
-# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
-# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
-# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
-# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
-# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
-# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
-# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
-# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
-# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
-# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
-# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
-# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
-# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
-# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
-# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
-# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
-# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
-# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
-# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
-# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
-# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
-# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
-# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
-# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
-# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
-# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
-# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
-# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
-# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
-# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
-# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
-# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
-# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
-# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
-# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
-# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
-# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
-# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
-# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
-# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
-# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
-# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
-# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
-# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
-# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
-# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
-# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
-# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
-# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
-# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
-# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
-# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
-# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
-# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
-# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
-# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
-# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
-# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
-# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
-# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
-# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
-# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
-# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
-# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
-# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
-# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
-# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
-# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
-# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
-# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
-# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
-# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
-# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
-# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
-# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
-# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
-# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
-# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
-# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
-# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
-# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
-# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
-# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
-# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
-# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
-# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
-# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
-# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
-# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
-# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
-# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
-# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
-# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
-# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
-# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
-# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
-# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
-# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
-# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
-# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
-# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
-# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
-# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
-# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
-# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
-# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
-# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
-# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
-# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
-# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
-# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
-# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
-# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
-# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
-# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
-# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
-# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
-# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
-# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
-# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
-# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
-# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
-# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
-# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
-# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
-# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
-# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
-# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
-# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
-# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
-# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
-# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
-# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
-# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
-# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
-# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
-# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
-# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
-# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
-# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
-# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
-# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
-# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
-# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
-# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
-# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
-# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
-# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
-# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
-# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
-# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
-# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
-# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
-# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
-# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
-# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
-# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
-# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
-# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
-# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
-# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
-# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
-# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
-# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
-# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
-# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
-# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
-# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
-# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
-# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
-# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
-# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
-# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
-# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
-# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
-# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
-# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
-# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
-# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
-# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
-# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
-# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
-# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
-# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
-# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
-# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
-# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
-# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
-# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
-# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
-# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
-# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
-# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
-# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
-# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
-# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
-# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
-# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
-# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
-# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
-# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
-# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
-# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
-# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
-# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
-# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
-# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
-# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
-# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
-# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
-# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
-# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
-# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
-# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
-# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
-# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
-# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
-# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
-# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
-# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
-# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
-# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
-# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
-# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
-# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
-# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
-# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
-# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
-# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
-# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
-# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
-# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
-# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
-# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
-# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
-# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
-# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
-# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
-# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
-# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
-# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
-# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
-# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
-# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
-# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
-# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
-# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
-# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
-# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
-# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
-# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
-# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
-# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
-# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
-# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
-# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
-# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
-# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
-# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
-# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
-# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
-# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
-# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
-# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
-# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
-# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
-# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
-# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
-# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
-# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
-# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
-# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
-# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
-# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
-# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
-# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
-# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
-# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
-# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
-# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
-# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
-# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
-# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
-# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
-# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
-# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
-# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
-# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
-# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
-# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
-# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
-# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
-# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
-# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
-# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
-# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
-# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
-# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
-# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
-# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
-# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
-# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
-# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
-# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
-# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
-# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
-# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
-# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
-# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
-# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
-# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
-# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
-# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
-# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
-# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
-# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
-# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
-# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
-# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
-# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
-# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
-# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
-# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
-# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
-# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
-# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
-# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
-# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
-# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
-# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
-# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
-# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
-# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
-# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
-# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
-# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
-# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
-# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
-# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
-# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
-# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
-# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
-# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
-# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
-# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
-# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
-# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
-# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
-# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
-# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
-# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
-# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
-# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
-# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
-# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
-# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
-# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
-# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
-# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
-# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
-# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
-# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
-# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
-# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
-# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
-# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
-# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
-# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
-# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
-# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
-# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
-# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
-# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
-# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
-# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
-# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
-# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
-# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
-# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
-# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
-# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
-# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
-# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
-# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
-# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
-# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
-# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
-# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
-# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
-# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
-# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
-# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
-# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
-# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
-# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
-# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
-# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
-# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
-# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
-# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
-# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
-# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
-# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
-# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
-# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
-# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
-# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
-# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
-# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
-# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
-# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
-# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
-# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
-# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
-# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
-# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
-# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
-# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
-# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
-# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
-# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
-# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
-# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
-# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
-# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
-# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
-# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
-# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
-# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
-# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
-# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
-# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
-# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
-# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
-# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
-# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
-# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
-# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
-# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
-# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
-# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
-# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
-# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
-# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
-# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
-# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
-# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
-# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
-# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
-# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
-# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
-# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
-# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
-# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
-# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
-# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
-# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
-# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
-# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
-# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
-# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
-# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
-# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
-# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
-# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
-# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
-# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
-# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
-# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
-# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
-# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
-# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
-# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
-# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
-# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
-# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
-# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
-# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
-# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
-# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
-# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
-# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
-# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
-# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
-# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
-# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
-# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
-# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
-# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
-# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
-# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
-# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
-# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
-# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
-# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
-# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
-# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
-# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
-# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
-# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
-# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
-# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
-# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
-# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
-# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
-# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
-# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
-# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
-# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
-# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
-# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
-# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
-# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
-# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
-# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
-# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
-# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
-# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
-# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
-# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
-# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
-# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
-# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
-# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
-# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
-# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
-# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
-# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
-# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
-# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
-# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
-# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
-# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
-# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
-# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
-# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
-# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
-# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
-# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
-# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
-# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
-# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
-# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
-# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
-# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
-# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
-# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
-# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
-# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
-# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
-# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
-# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
-# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
-# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
-# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
-# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
-# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
-# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
-# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
-# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
-# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
-# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
-# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
-# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
-# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
-# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
-# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
-# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
-# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
-# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
-# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
-# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
-# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
-# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
-# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
-# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
-# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
-# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
-# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
-# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
-# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
-# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
-# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
-# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
-# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
-# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
-# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
-# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
-# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
-# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
-# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
-# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
-# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
-# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
-# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
-# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
-# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
-# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
-# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
-# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
-# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
-# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
-# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
-# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
-# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
-# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
-# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
-# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
-# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
-# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
-# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
-# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
-# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
-# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
-# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
-# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
-# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
-# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
-# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
-# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
-# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
-# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
-# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
-# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
-# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
-# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
-# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
-# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
-# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
-# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
-# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
-# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
-# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
-# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
-# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
-# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
-# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
-# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
-# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
-# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
-# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
-# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
-# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
-# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
-# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
-# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
-# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
-# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
-# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
-# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
-# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
-# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
-# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
-# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
-# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
-# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
-# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
-# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
-# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
-# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
-# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
-# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
-# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
-# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
-# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
-# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
-# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
-# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
-# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
-# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
-# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
-# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
-# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
-# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
-# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
-# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
-# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
-# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
-# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
-# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
-# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
-# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
-# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
-# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
-# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
-# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
-# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
-# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
-# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
-# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
-# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
-# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
-# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
-# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
-# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
-# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
-# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
-# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
-# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
-# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
-# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
-# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
-# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
-# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
-# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
-# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
-# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
-# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
-# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
-# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
-# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
-# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
-# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
-# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
-# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
-# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
-# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
-# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
-# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
-# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
-# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
-# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
-# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
-# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
-# 11:38:37 --- INFO: Training process end.
-# 11:38:37 --- INFO: Evaluating process start.
-# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
-# 11:38:37 --- INFO: Evaluating process end.
-# 11:38:37 --- INFO: Predicting process start.
-# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
-# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
-# 11:38:38 --- INFO: Predicting process end.
-# 11:38:38 --- INFO: --------------------- GNN end ----------------------
-# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
-# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
-# 11:38:38 --- INFO: Loading dataset.
-# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:38:39 --- INFO: Calculating NTScore for each niche.
-# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
-# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
-# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
-# 11:38:39 --- INFO: Output NTScore tables.
-# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
+
+source ~/.bash_profile
+conda activate ONTraC_v2_py311
+ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
+# ##################################################################################
+#
+# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
+# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
+# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
+# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
+# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
+#
+# version: 2.0a11
+#
+# ##################################################################################
+# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
+# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
+# 11:33:23 --- INFO: -------- I/O options -------
+# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
+# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
+# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
+# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
+# 11:33:23 --- INFO: -------- preprocessing options -------
+# 11:33:23 --- INFO: resolution: 10.0
+# 11:33:23 --- INFO: -------- niche net constr options -------
+# 11:33:23 --- INFO: n_cpu: 4
+# 11:33:23 --- INFO: n_neighbors: 50
+# 11:33:23 --- INFO: n_local: 20
+# 11:33:23 --- INFO: embedding_adjust: False
+# 11:33:23 --- INFO: sigma: 1
+# 11:33:23 --- INFO: -------- train options -------
+# 11:33:23 --- INFO: device: cpu
+# 11:33:23 --- INFO: epochs: 1000
+# 11:33:23 --- INFO: batch_size: 0
+# 11:33:23 --- INFO: patience: 100
+# 11:33:23 --- INFO: min_delta: 0.001
+# 11:33:23 --- INFO: min_epochs: 50
+# 11:33:23 --- INFO: seed: 42
+# 11:33:23 --- INFO: lr: 0.03
+# 11:33:23 --- INFO: hidden_feats: 4
+# 11:33:23 --- INFO: k: 6
+# 11:33:23 --- INFO: modularity_loss_weight: 1.0
+# 11:33:23 --- INFO: purity_loss_weight: 100.0
+# 11:33:23 --- INFO: regularization_loss_weight: 0.1
+# 11:33:23 --- INFO: beta: 0.3
+# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
+# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
+# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
+# 11:33:23 --- INFO: ------------- Niche network construct ---------------
+# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
+# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
+# 11:33:28 --- INFO: Generating samples.yaml file.
+# 11:33:28 --- INFO: ------------ Niche network construct end ------------
+# 11:33:28 --- INFO: ------------------------ GNN ------------------------
+# 11:33:28 --- INFO: Loading dataset.
+# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
+# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
+# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
+# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
+# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
+# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
+# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
+# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
+# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
+# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
+# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
+# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
+# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
+# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
+# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
+# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
+# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
+# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
+# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
+# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
+# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
+# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
+# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
+# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
+# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
+# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
+# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
+# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
+# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
+# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
+# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
+# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
+# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
+# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
+# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
+# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
+# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
+# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
+# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
+# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
+# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
+# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
+# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
+# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
+# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
+# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
+# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
+# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
+# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
+# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
+# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
+# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
+# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
+# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
+# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
+# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
+# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
+# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
+# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
+# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
+# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
+# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
+# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
+# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
+# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
+# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
+# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
+# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
+# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
+# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
+# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
+# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
+# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
+# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
+# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
+# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
+# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
+# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
+# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
+# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
+# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
+# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
+# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
+# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
+# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
+# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
+# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
+# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
+# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
+# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
+# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
+# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
+# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
+# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
+# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
+# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
+# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
+# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
+# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
+# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
+# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
+# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
+# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
+# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
+# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
+# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
+# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
+# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
+# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
+# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
+# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
+# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
+# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
+# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
+# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
+# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
+# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
+# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
+# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
+# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
+# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
+# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
+# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
+# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
+# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
+# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
+# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
+# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
+# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
+# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
+# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
+# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
+# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
+# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
+# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
+# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
+# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
+# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
+# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
+# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
+# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
+# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
+# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
+# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
+# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
+# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
+# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
+# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
+# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
+# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
+# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
+# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
+# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
+# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
+# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
+# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
+# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
+# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
+# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
+# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
+# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
+# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
+# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
+# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
+# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
+# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
+# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
+# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
+# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
+# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
+# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
+# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
+# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
+# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
+# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
+# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
+# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
+# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
+# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
+# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
+# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
+# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
+# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
+# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
+# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
+# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
+# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
+# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
+# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
+# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
+# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
+# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
+# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
+# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
+# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
+# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
+# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
+# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
+# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
+# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
+# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
+# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
+# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
+# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
+# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
+# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
+# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
+# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
+# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
+# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
+# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
+# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
+# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
+# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
+# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
+# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
+# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
+# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
+# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
+# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
+# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
+# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
+# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
+# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
+# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
+# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
+# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
+# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
+# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
+# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
+# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
+# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
+# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
+# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
+# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
+# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
+# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
+# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
+# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
+# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
+# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
+# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
+# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
+# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
+# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
+# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
+# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
+# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
+# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
+# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
+# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
+# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
+# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
+# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
+# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
+# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
+# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
+# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
+# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
+# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
+# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
+# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
+# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
+# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
+# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
+# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
+# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
+# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
+# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
+# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
+# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
+# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
+# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
+# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
+# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
+# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
+# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
+# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
+# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
+# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
+# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
+# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
+# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
+# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
+# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
+# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
+# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
+# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
+# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
+# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
+# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
+# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
+# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
+# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
+# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
+# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
+# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
+# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
+# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
+# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
+# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
+# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
+# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
+# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
+# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
+# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
+# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
+# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
+# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
+# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
+# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
+# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
+# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
+# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
+# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
+# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
+# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
+# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
+# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
+# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
+# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
+# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
+# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
+# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
+# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
+# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
+# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
+# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
+# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
+# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
+# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
+# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
+# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
+# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
+# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
+# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
+# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
+# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
+# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
+# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
+# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
+# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
+# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
+# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
+# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
+# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
+# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
+# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
+# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
+# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
+# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
+# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
+# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
+# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
+# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
+# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
+# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
+# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
+# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
+# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
+# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
+# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
+# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
+# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
+# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
+# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
+# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
+# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
+# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
+# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
+# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
+# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
+# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
+# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
+# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
+# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
+# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
+# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
+# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
+# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
+# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
+# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
+# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
+# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
+# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
+# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
+# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
+# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
+# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
+# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
+# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
+# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
+# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
+# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
+# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
+# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
+# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
+# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
+# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
+# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
+# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
+# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
+# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
+# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
+# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
+# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
+# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
+# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
+# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
+# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
+# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
+# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
+# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
+# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
+# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
+# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
+# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
+# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
+# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
+# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
+# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
+# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
+# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
+# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
+# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
+# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
+# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
+# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
+# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
+# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
+# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
+# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
+# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
+# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
+# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
+# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
+# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
+# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
+# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
+# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
+# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
+# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
+# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
+# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
+# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
+# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
+# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
+# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
+# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
+# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
+# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
+# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
+# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
+# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
+# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
+# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
+# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
+# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
+# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
+# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
+# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
+# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
+# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
+# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
+# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
+# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
+# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
+# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
+# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
+# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
+# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
+# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
+# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
+# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
+# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
+# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
+# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
+# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
+# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
+# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
+# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
+# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
+# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
+# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
+# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
+# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
+# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
+# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
+# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
+# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
+# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
+# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
+# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
+# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
+# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
+# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
+# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
+# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
+# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
+# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
+# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
+# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
+# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
+# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
+# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
+# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
+# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
+# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
+# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
+# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
+# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
+# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
+# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
+# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
+# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
+# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
+# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
+# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
+# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
+# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
+# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
+# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
+# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
+# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
+# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
+# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
+# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
+# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
+# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
+# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
+# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
+# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
+# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
+# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
+# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
+# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
+# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
+# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
+# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
+# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
+# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
+# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
+# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
+# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
+# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
+# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
+# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
+# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
+# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
+# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
+# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
+# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
+# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
+# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
+# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
+# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
+# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
+# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
+# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
+# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
+# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
+# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
+# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
+# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
+# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
+# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
+# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
+# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
+# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
+# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
+# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
+# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
+# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
+# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
+# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
+# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
+# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
+# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
+# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
+# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
+# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
+# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
+# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
+# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
+# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
+# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
+# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
+# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
+# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
+# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
+# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
+# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
+# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
+# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
+# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
+# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
+# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
+# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
+# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
+# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
+# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
+# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
+# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
+# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
+# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
+# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
+# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
+# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
+# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
+# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
+# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
+# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
+# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
+# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
+# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
+# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
+# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
+# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
+# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
+# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
+# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
+# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
+# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
+# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
+# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
+# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
+# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
+# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
+# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
+# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
+# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
+# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
+# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
+# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
+# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
+# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
+# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
+# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
+# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
+# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
+# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
+# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
+# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
+# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
+# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
+# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
+# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
+# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
+# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
+# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
+# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
+# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
+# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
+# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
+# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
+# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
+# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
+# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
+# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
+# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
+# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
+# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
+# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
+# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
+# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
+# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
+# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
+# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
+# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
+# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
+# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
+# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
+# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
+# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
+# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
+# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
+# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
+# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
+# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
+# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
+# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
+# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
+# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
+# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
+# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
+# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
+# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
+# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
+# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
+# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
+# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
+# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
+# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
+# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
+# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
+# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
+# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
+# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
+# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
+# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
+# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
+# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
+# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
+# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
+# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
+# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
+# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
+# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
+# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
+# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
+# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
+# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
+# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
+# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
+# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
+# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
+# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
+# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
+# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
+# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
+# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
+# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
+# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
+# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
+# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
+# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
+# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
+# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
+# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
+# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
+# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
+# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
+# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
+# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
+# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
+# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
+# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
+# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
+# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
+# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
+# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
+# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
+# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
+# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
+# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
+# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
+# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
+# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
+# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
+# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
+# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
+# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
+# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
+# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
+# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
+# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
+# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
+# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
+# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
+# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
+# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
+# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
+# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
+# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
+# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
+# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
+# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
+# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
+# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
+# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
+# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
+# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
+# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
+# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
+# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
+# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
+# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
+# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
+# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
+# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
+# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
+# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
+# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
+# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
+# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
+# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
+# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
+# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
+# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
+# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
+# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
+# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
+# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
+# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
+# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
+# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
+# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
+# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
+# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
+# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
+# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
+# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
+# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
+# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
+# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
+# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
+# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
+# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
+# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
+# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
+# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
+# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
+# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
+# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
+# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
+# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
+# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
+# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
+# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
+# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
+# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
+# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
+# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
+# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
+# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
+# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
+# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
+# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
+# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
+# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
+# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
+# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
+# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
+# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
+# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
+# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
+# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
+# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
+# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
+# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
+# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
+# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
+# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
+# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
+# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
+# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
+# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
+# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
+# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
+# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
+# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
+# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
+# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
+# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
+# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
+# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
+# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
+# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
+# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
+# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
+# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
+# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
+# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
+# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
+# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
+# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
+# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
+# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
+# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
+# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
+# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
+# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
+# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
+# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
+# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
+# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
+# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
+# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
+# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
+# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
+# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
+# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
+# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
+# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
+# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
+# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
+# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
+# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
+# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
+# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
+# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
+# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
+# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
+# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
+# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
+# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
+# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
+# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
+# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
+# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
+# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
+# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
+# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
+# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
+# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
+# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
+# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
+# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
+# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
+# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
+# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
+# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
+# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
+# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
+# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
+# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
+# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
+# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
+# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
+# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
+# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
+# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
+# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
+# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
+# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
+# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
+# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
+# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
+# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
+# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
+# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
+# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
+# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
+# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
+# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
+# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
+# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
+# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
+# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
+# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
+# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
+# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
+# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
+# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
+# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
+# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
+# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
+# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
+# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
+# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
+# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
+# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
+# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
+# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
+# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
+# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
+# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
+# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
+# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
+# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
+# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
+# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
+# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
+# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
+# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
+# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
+# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
+# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
+# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
+# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
+# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
+# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
+# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
+# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
+# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
+# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
+# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
+# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
+# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
+# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
+# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
+# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
+# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
+# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
+# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
+# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
+# 11:38:37 --- INFO: Training process end.
+# 11:38:37 --- INFO: Evaluating process start.
+# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
+# 11:38:37 --- INFO: Evaluating process end.
+# 11:38:37 --- INFO: Predicting process start.
+# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
+# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
+# 11:38:38 --- INFO: Predicting process end.
+# 11:38:38 --- INFO: --------------------- GNN end ----------------------
+# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
+# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
+# 11:38:38 --- INFO: Loading dataset.
+# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:38:39 --- INFO: Calculating NTScore for each niche.
+# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
+# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
+# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
+# 11:38:39 --- INFO: Output NTScore tables.
+# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
16.1.3 visualization
16.1.3.1 load ONTraC results
-
+
The NTScore and binarized niche cluster info were stored in cell metadata
-
-# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
-# <char> <char> <char> <char> <char> <char> <char> <int> <num>
-# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
-# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
-# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
-# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
-# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
-# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
+
+# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
+# <char> <char> <char> <char> <char> <char> <char> <int> <num>
+# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
+# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
+# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
+# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
+# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
+# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
The probability matrix of each cell assigned to each niche cluster and
connectivity between niche cluster were stored here.
-
-
+
+
16.1.3.2 niche cluster prob
-spatFeatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'niche cluster',
- expression_values = "prob",
- group_by = 'list_ID',
- feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
- point_border_col = 'gray'
- )
+spatFeatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'niche cluster',
+ expression_values = "prob",
+ group_by = 'list_ID',
+ feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
+ point_border_col = 'gray'
+ )
16.1.3.3 binarized niche cluster
-spatPlot2D(giotto_obj,
- spat_unit = "cell",
- group_by = 'list_ID',
- cell_color = "NicheCluster",
- color_as_factor = T,
- point_size = 1,
- point_border_stroke = NA,
- title="Niche cluster")
+spatPlot2D(giotto_obj,
+ spat_unit = "cell",
+ group_by = 'list_ID',
+ cell_color = "NicheCluster",
+ color_as_factor = T,
+ point_size = 1,
+ point_border_stroke = NA,
+ title="Niche cluster")
16.1.3.5 NT score
-spatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'rna',
- group_by = 'list_ID',
- cell_color = "NTScore",
- color_as_factor = FALSE,
- cell_color_gradient = "turbo",
- point_size = 1,
- point_border_stroke = NA
- )
+spatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ group_by = 'list_ID',
+ cell_color = "NTScore",
+ color_as_factor = FALSE,
+ cell_color_gradient = "turbo",
+ point_size = 1,
+ point_border_stroke = NA
+ )
-
+
diff --git a/interoperability-with-other-frameworks.html b/interoperability-with-other-frameworks.html
index 7050533..347d3af 100644
--- a/interoperability-with-other-frameworks.html
+++ b/interoperability-with-other-frameworks.html
@@ -521,10 +521,10 @@ 15 Interoperability with other fr
15.1 Load Giotto object
To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process:
-
+
Load a Giotto mini Visium object, which will be used for demonstrating interoperability.
-
+
15.2 Seurat
@@ -533,99 +533,99 @@ 15.2 Seurat
15.2.1 Conversion of Giotto Object to Seurat Object
To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function
-
+
Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data.
-gToS <- giottoToSeuratV5(gobject = gobject,
- spat_unit = "cell")
-
-plot1 <- VlnPlot(gToS,
- features = "nCount_rna",
- pt.size = 0.1) +
- NoLegend()
-
-plot2 <- SpatialFeaturePlot(gToS,
- features = "nCount_rna",
- pt.size.factor = 2) +
- theme(legend.position = "right")
-
-wrap_plots(plot1, plot2)
+gToS <- giottoToSeuratV5(gobject = gobject,
+ spat_unit = "cell")
+
+plot1 <- VlnPlot(gToS,
+ features = "nCount_rna",
+ pt.size = 0.1) +
+ NoLegend()
+
+plot2 <- SpatialFeaturePlot(gToS,
+ features = "nCount_rna",
+ pt.size.factor = 2) +
+ theme(legend.position = "right")
+
+wrap_plots(plot1, plot2)
15.2.1.1 Apply SCTransform
We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function.
-
+
15.2.1.2 Dimension Reduction
We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object:
-
+
15.2.2 Conversion of Seurat object Back to Giotto Object
To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks.
-giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
- spatial_assay = "rna",
- dim_reduction = c("pca", "umap"),
- sp_network = "Delaunay_network",
- nn_network = c("sNN.pca", "custom_NN" ))
+giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
+ spatial_assay = "rna",
+ dim_reduction = c("pca", "umap"),
+ sp_network = "Delaunay_network",
+ nn_network = c("sNN.pca", "custom_NN" ))
15.2.2.1 Clustering and Plotting UMAP
Here we perform K-means clustering on the UMAP results obtained from the Seurat object:
-## k-means clustering
-giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
- dim_reduction_to_use = "pca")
-
-#Plot UMAP post-clustering to visualize kmeans
-graph2 <- Giotto::plotUMAP(
- gobject = giottoFromSeurat,
- cell_color = "kmeans",
- show_NN_network = TRUE,
- point_size = 2.5
-)
+## k-means clustering
+giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
+ dim_reduction_to_use = "pca")
+
+#Plot UMAP post-clustering to visualize kmeans
+graph2 <- Giotto::plotUMAP(
+ gobject = giottoFromSeurat,
+ cell_color = "kmeans",
+ show_NN_network = TRUE,
+ point_size = 2.5
+)
15.2.2.2 Spatial CoExpression
We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function:
-ranktest <- binSpect(giottoFromSeurat,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "Delaunay_network")
-
-ext_spatial_genes <- ranktest[1:300,]$feats
-
-spat_cor_netw_DT <- detectSpatialCorFeats(
- giottoFromSeurat,
- method = "network",
- spatial_network_name = "Delaunay_network",
- subset_feats = ext_spatial_genes)
-
-top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
- feats = "Dsp",
- show_top_feats = 10)
-
-spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
- name = "spat_netw_clus",
- k = 7)
-heatmSpatialCorFeats(
- giottoFromSeurat,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL),
- save_plot = TRUE,
- show_plot = TRUE,
- return_plot = FALSE,
- save_param = list(base_height = 6, base_width = 8, units = 'cm'))
+ranktest <- binSpect(giottoFromSeurat,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "Delaunay_network")
+
+ext_spatial_genes <- ranktest[1:300,]$feats
+
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ giottoFromSeurat,
+ method = "network",
+ spatial_network_name = "Delaunay_network",
+ subset_feats = ext_spatial_genes)
+
+top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
+ feats = "Dsp",
+ show_top_feats = 10)
+
+spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
+ name = "spat_netw_clus",
+ k = 7)
+heatmSpatialCorFeats(
+ giottoFromSeurat,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ return_plot = FALSE,
+ save_param = list(base_height = 6, base_width = 8, units = 'cm'))
@@ -635,60 +635,60 @@ 15.3 SpatialExperiment
To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries.
-library(SpatialExperiment)
-library(ggspavis)
-library(pheatmap)
-library(scater)
-library(scran)
-library(nnSVG)
+library(SpatialExperiment)
+library(ggspavis)
+library(pheatmap)
+library(scater)
+library(scran)
+library(nnSVG)
15.3.1 Convert Giotto Object to SpatialExperiment Object
To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function.
-
+
The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use:
-
+
15.3.1.1 Identify top spatially variable genes with nnSVG
We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate:
-# One of the assays should be "logcounts"
-# We rename the normalized assay to "logcounts"
-assayNames(spe)[[2]] <- "logcounts"
-
-# Create model matrix for leiden clustering labels
-X <- model.matrix(~ colData(spe)$leiden_clus)
-dim(X)
+# One of the assays should be "logcounts"
+# We rename the normalized assay to "logcounts"
+assayNames(spe)[[2]] <- "logcounts"
+
+# Create model matrix for leiden clustering labels
+X <- model.matrix(~ colData(spe)$leiden_clus)
+dim(X)
- Run nnSVG
This step will take several minutes to run
-
+
15.3.2 Conversion of SpatialExperiment object back to Giotto
We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object.
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-print(giottoFromSPE)
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+print(giottoFromSPE)
15.3.2.1 Plotting top genes from nnSVG in Giotto
Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object.
-ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
- output = "data.table")
-
-ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
-spatFeatPlot2D(giottoFromSPE,
- expression_values = "scaled_rna_cell",
- feats = ext_spatial_genes[1:4],
- point_size = 2)
+ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
+ output = "data.table")
+
+ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
+spatFeatPlot2D(giottoFromSPE,
+ expression_values = "scaled_rna_cell",
+ feats = ext_spatial_genes[1:4],
+ point_size = 2)
@@ -701,97 +701,97 @@ 15.4 AnnData
15.4.1 Load Required Libraries
To start, we need to load the necessary libraries, including reticulate for interfacing with Python.
-
+
15.4.2 Specify Path for Results
First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions.
-# Specify path to which results may be saved
-results_directory <- "results/03_session4/giotto_anndata_conversion/"
-
-instrs <- showGiottoInstructions(gobject)
-
-mini_gobject <- replaceGiottoInstructions(gobject = gobject,
- instructions = instrs)
+# Specify path to which results may be saved
+results_directory <- "results/03_session4/giotto_anndata_conversion/"
+
+instrs <- showGiottoInstructions(gobject)
+
+mini_gobject <- replaceGiottoInstructions(gobject = gobject,
+ instructions = instrs)
15.4.2.1 Create Default kNN Network
We will create a k-nearest neighbor (kNN) network using mostly default parameters.
-
+
15.4.3 Giotto To AnnData
To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData()
-
+
Next, we import scanpy and perform a series of preprocessing steps on the AnnData object.
-scanpy <- import("scanpy")
-
-adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
-
-# Normalize total counts per cell
-scanpy$pp$normalize_total(adata, target_sum=1e4)
-
-# Log-transform the data
-scanpy$pp$log1p(adata)
-
-# Perform PCA
-scanpy$pp$pca(adata, n_comps=40L)
-
-# Compute the neighborhood graph
-scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
-
-# Run UMAP
-scanpy$tl$umap(adata)
-
-# Save the processed AnnData object
-adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
-
-processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
+scanpy <- import("scanpy")
+
+adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
+
+# Normalize total counts per cell
+scanpy$pp$normalize_total(adata, target_sum=1e4)
+
+# Log-transform the data
+scanpy$pp$log1p(adata)
+
+# Perform PCA
+scanpy$pp$pca(adata, n_comps=40L)
+
+# Compute the neighborhood graph
+scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
+
+# Run UMAP
+scanpy$tl$umap(adata)
+
+# Save the processed AnnData object
+adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
+
+processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
15.4.4 Convert AnnData to Giotto
Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto.
-
+
15.4.4.1 UMAP Visualization
Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object.
-Giotto::plotUMAP(
- gobject = giottoFromAnndata,
- dim_reduction_name = "umap.ad",
- cell_color = "leiden_clus",
- point_size = 3
-)
+Giotto::plotUMAP(
+ gobject = giottoFromAnndata,
+ dim_reduction_name = "umap.ad",
+ cell_color = "leiden_clus",
+ point_size = 3
+)
15.5 Create mini Vizgen object
-mini_gobject <- loadGiottoMini(dataset = "vizgen",
- python_path = NULL)
-
-mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
- instructions = instrs)
-mini_gobject <- createNearestNetwork(gobject = mini_gobject,
- spat_unit = "aggregate",
- feat_type = "rna",
- type = "kNN",
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap",
- k = 6,
- name = "new_network")
+mini_gobject <- loadGiottoMini(dataset = "vizgen",
+ python_path = NULL)
+
+mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
+ instructions = instrs)
+mini_gobject <- createNearestNetwork(gobject = mini_gobject,
+ spat_unit = "aggregate",
+ feat_type = "rna",
+ type = "kNN",
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap",
+ k = 6,
+ name = "new_network")
Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details.
-
+
diff --git a/introduction-to-the-giotto-package.html b/introduction-to-the-giotto-package.html
index e22d6d1..0ab2014 100644
--- a/introduction-to-the-giotto-package.html
+++ b/introduction-to-the-giotto-package.html
@@ -546,19 +546,58 @@ 4.3 Installation + python environ
4.3.1 Giotto installation
Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository.
Much of this already covered in Section 2.2, but the highlights are:
-
-- Ensure your system
-
-To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak.
+
+4.3.1.1 System prerequisites
+
+- for windows, Rtools needs to be installed
+- a major dependency terra needs
GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux
+
+
+
+4.3.1.2 Installation of released version
+To install the currently released version of Giotto in a single step:
+This should automatically install all the Giotto dependencies and other Giotto module packages.
+
+
+4.3.1.3 Installation of dev branch Giotto packages
+You can install dev branch versions by using devtools::install_github()
+Core module dev branchs:
+
+- “drieslab/Giotto@suite_dev”
+- “drieslab/GiottoVisuals@dev”
+- “drieslab/GiottoClass@dev”
+- “drieslab/GiottoUtils@dev”
+
+
+pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages.
+
+
+4.3.1.4 Common install issues
+If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work.
+
+If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it.
+
+## # Downloading packages -------------------------------------------------------
+## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s]
+## Successfully downloaded 1 package in 1.2 seconds.
+##
+## The following package(s) will be installed:
+## - irlba [2.3.5.1]
+## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu".
+##
+## # Installing packages --------------------------------------------------------
+## - Installing irlba ... OK [built from source and cached in 5.7s]
+## Successfully installed 1 package in 5.7 seconds.
+
4.3.2 Python environment
4.3.2.1 Default installation
In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment.
-
+
Two things are needed for python to work:
- A conda (e.g. miniconda or anaconda) installation which is the package and environment management system.
@@ -580,17 +619,17 @@ 4.3.2.1 Default installation
4.3.2.2 Custom installs
Custom python environments can be made by first setting up a new environment and establishing the name and python version to use.
-
+
Following that, one or more python packages to install can be added to the environment.
-reticulate::py_install(
- pip = TRUE,
- envname = '[name of env]',
- packages = c(
- "package1",
- "package2",
- "..."
- )
-)
+reticulate::py_install(
+ pip = TRUE,
+ envname = '[name of env]',
+ packages = c(
+ "package1",
+ "package2",
+ "..."
+ )
+)
Once an environment has been set up, Giotto can hook into it.
@@ -612,17 +651,56 @@ 4.3.2.3 Using a specific environm
Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto.
You would run file.edit("~/.Rprofile") and then add options("giotto.py_path" = "[path to env or envname]") as a line so that it is automatically set at the start of each session.
If a specific environment should only be used a couple times then method 1 is easiest:
-
+
To check which conda environments exist on your machine:
-
+
Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running:
-
+
4.4 Giotto instructions
-Giotto
+Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are:
+
+- python_path - when set, will activate a python environment
+- save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render.
+- save_plot - whether to save plots to the
save_dir
+- return_plot - whether to return the plot objects. When FALSE, only NULL is returned
+- show_plot - whether to show the plot in the viewer
+
+These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function.
+library(Giotto)
+save_dir <- "results/01_session2/"
+
+# this call will also intialize the python env
+instrs <- createGiottoInstructions(
+ save_dir = save_dir, # working directory is the default
+ show_plot = FALSE,
+ save_plot = TRUE,
+ return_plot = FALSE,
+ python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default
+)
+force(instrs)
+Giotto object creation functions all have an instructions param for passing in instructions objects.
+giotto objects will also respond to the instructions() generic.
+test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object
+
+# example plot
+g <- GiottoData::loadGiottoMini("visium")
+instructions(g) <- instrs
+instructions(g, "show_plot") # instructions say not to plot to viewer
+spatPlot2D(g, show_image = TRUE, image_name = "image")
+# instead it will directly write to the results folder
+As an example, you can also set individual instructions
+
+
+
+
+Figure 4.2: example image output
+
+
diff --git a/multi-omics-integration.html b/multi-omics-integration.html
index 47cad23..e76e25a 100644
--- a/multi-omics-integration.html
+++ b/multi-omics-integration.html
@@ -520,14 +520,14 @@ 14 Multi-omics integration
14.1 The CytAssist technology
The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image.
-
+
Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics.
The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument.
-
+
Figure 14.2: CytAssist workflow. Source: 10X genomics.
@@ -542,19 +542,19 @@
14.2 Introduction to the spatial
14.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/03_session3")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
+dir.create("data/03_session3")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”.
-
+
14.4 Create the Giotto object
@@ -564,124 +564,124 @@ 14.4 Create the Giotto objectx,y(,z) coordinates for cells or spots (or the path to)
createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/03_session3/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-# Provide the path to the data folder
-data_path <- "data/03_session3/"
-
-# Create object directly from the data folder
-visium_tonsil <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/03_session3/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+# Provide the path to the data folder
+data_path <- "data/03_session3/"
+
+# Create object directly from the data folder
+visium_tonsil <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
- Print the information of the object, note that both rna and protein are listed in the expression slot.
-
+
14.5 Subset on spots that were covered by tissue
-spatPlot2D(
- gobject = visium_tonsil,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE,
- image_name = "image"
-)
-
+spatPlot2D(
+ gobject = visium_tonsil,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE,
+ image_name = "image"
+)
+
Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object.
-
+
14.6 RNA processing
- Run the Filtering, normalization, and statistics steps using only the RNA feature.
-visium_tonsil <- filterGiotto(
- gobject = visium_tonsil,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil)
+visium_tonsil <- filterGiotto(
+ gobject = visium_tonsil,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1000,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil)
- Dimension reduction
Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs.
-visium_tonsil <- calculateHVF(gobject = visium_tonsil)
-
-visium_tonsil <- runPCA(gobject = visium_tonsil)
+visium_tonsil <- calculateHVF(gobject = visium_tonsil)
+
+visium_tonsil <- runPCA(gobject = visium_tonsil)
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the RNA-based Leiden clusters.
-
+
- Visualization
Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot.
-plotUMAP(gobject = visium_tonsil,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters.
Plot the spatial distribution of the RNA-based Leiden cluster per spot.
-spatPlot2D(gobject = visium_tonsil,
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.5: Spatial distribution of RNA-based Leiden clusters.
@@ -693,80 +693,80 @@
14.7 Protein processingvisium_tonsil <- filterGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein")
+visium_tonsil <- filterGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein")
- Dimension reduction
Calculate the principal components using all the proteins available in the dataset.
-
+
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the Protein-based Leiden clusters.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- resolution = 1,
- n_iterations = 1000)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ resolution = 1,
+ n_iterations = 1000)
- Visualization
Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters.
Plot the spatial distribution of the Protein-based Leiden clusters.
-spatPlot2D(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.7: Spatial distribution of Protein-based Leiden clusters.
@@ -779,68 +779,68 @@
14.8 Multi-omics integrationCalculate kNN
Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type.
-## RNA modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
-
-## Protein modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
+## RNA modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
+
+## Protein modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
- Run WNN
Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object.
-visium_tonsil <- runWNN(visium_tonsil,
- spat_unit = "cell",
- modality_1 = "rna",
- modality_2 = "protein",
- pca_name_modality_1 = "pca",
- pca_name_modality_2 = "protein.pca",
- k = 20,
- integrated_feat_type = NULL,
- matrix_result_name = NULL,
- w_name_modality_1 = NULL,
- w_name_modality_2 = NULL,
- verbose = TRUE)
+visium_tonsil <- runWNN(visium_tonsil,
+ spat_unit = "cell",
+ modality_1 = "rna",
+ modality_2 = "protein",
+ pca_name_modality_1 = "pca",
+ pca_name_modality_2 = "protein.pca",
+ k = 20,
+ integrated_feat_type = NULL,
+ matrix_result_name = NULL,
+ w_name_modality_1 = NULL,
+ w_name_modality_2 = NULL,
+ verbose = TRUE)
- Run Integrated umap
Calculate the UMAP using the weights of each feature per spot.
-visium_tonsil <- runIntegratedUMAP(visium_tonsil,
- modality1 = "rna",
- modality2 = "protein",
- spread = 7,
- min_dist = 1,
- force = FALSE)
+visium_tonsil <- runIntegratedUMAP(visium_tonsil,
+ modality1 = "rna",
+ modality2 = "protein",
+ spread = 7,
+ min_dist = 1,
+ force = FALSE)
- Calculate integrated clusters
Calculate the multiomics-based Leiden clusters using the weights of each feature per spot.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- nn_network_to_use = "kNN",
- network_name = "integrated_kNN",
- name = "integrated_leiden_clus",
- resolution = 1)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ nn_network_to_use = "kNN",
+ network_name = "integrated_kNN",
+ name = "integrated_leiden_clus",
+ resolution = 1)
- Visualize the integrated UMAP
Plot the integrated UMAP and color the spots using the integrated Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- dim_reduction_name = "integrated.umap",
- point_size = 2,
- title = "Integrated UMAP using Integrated Leiden clusters")
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ dim_reduction_name = "integrated.umap",
+ point_size = 2,
+ title = "Integrated UMAP using Integrated Leiden clusters")
+
Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters.
@@ -850,14 +850,14 @@
14.8 Multi-omics integrationVisualize spatial plot with integrated clusters
Plot the spatial distribution of the integrated Leiden clusters.
-spatPlot2D(visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- point_size = 3,
- show_image = TRUE,
- title = "Integrated Leiden clustering")
-
+spatPlot2D(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ point_size = 3,
+ show_image = TRUE,
+ title = "Integrated Leiden clustering")
+
Figure 14.9: Spatial distribution of the integrated Leiden clusters.
@@ -866,85 +866,85 @@
14.8 Multi-omics integration
14.9 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
- [7] matrixStats_1.3.0 compiler_4.4.1
- [9] png_0.1-8 systemfonts_1.1.0
- [11] vctrs_0.6.5 reshape2_1.4.4
- [13] stringr_1.5.1 pkgconfig_2.0.3
- [15] SpatialExperiment_1.14.0 crayon_1.5.3
- [17] fastmap_1.2.0 backports_1.5.0
- [19] magick_2.8.4 XVector_0.44.0
- [21] labeling_0.4.3 utf8_1.2.4
- [23] rmarkdown_2.27 UCSC.utils_1.0.0
- [25] ragg_1.3.2 purrr_1.0.2
- [27] xfun_0.46 beachmat_2.20.0
- [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
- [31] jsonlite_1.8.8 DelayedArray_0.30.1
- [33] BiocParallel_1.38.0 terra_1.7-78
- [35] irlba_2.3.5.1 parallel_4.4.1
- [37] R6_2.5.1 stringi_1.8.4
- [39] RColorBrewer_1.1-3 reticulate_1.38.0
- [41] parallelly_1.37.1 GenomicRanges_1.56.1
- [43] scattermore_1.2 Rcpp_1.0.13
- [45] bookdown_0.40 SummarizedExperiment_1.34.0
- [47] knitr_1.48 future.apply_1.11.2
- [49] R.utils_2.12.3 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
+ [7] matrixStats_1.3.0 compiler_4.4.1
+ [9] png_0.1-8 systemfonts_1.1.0
+ [11] vctrs_0.6.5 reshape2_1.4.4
+ [13] stringr_1.5.1 pkgconfig_2.0.3
+ [15] SpatialExperiment_1.14.0 crayon_1.5.3
+ [17] fastmap_1.2.0 backports_1.5.0
+ [19] magick_2.8.4 XVector_0.44.0
+ [21] labeling_0.4.3 utf8_1.2.4
+ [23] rmarkdown_2.27 UCSC.utils_1.0.0
+ [25] ragg_1.3.2 purrr_1.0.2
+ [27] xfun_0.46 beachmat_2.20.0
+ [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
+ [31] jsonlite_1.8.8 DelayedArray_0.30.1
+ [33] BiocParallel_1.38.0 terra_1.7-78
+ [35] irlba_2.3.5.1 parallel_4.4.1
+ [37] R6_2.5.1 stringi_1.8.4
+ [39] RColorBrewer_1.1-3 reticulate_1.38.0
+ [41] parallelly_1.37.1 GenomicRanges_1.56.1
+ [43] scattermore_1.2 Rcpp_1.0.13
+ [45] bookdown_0.40 SummarizedExperiment_1.34.0
+ [47] knitr_1.48 future.apply_1.11.2
+ [49] R.utils_2.12.3 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/reference-keys.txt b/reference-keys.txt
index a17604d..555f956 100644
--- a/reference-keys.txt
+++ b/reference-keys.txt
@@ -4,115 +4,116 @@ fig:unnamed-chunk-14
fig:unnamed-chunk-15
fig:unnamed-chunk-16
fig:unnamed-chunk-18
-fig:unnamed-chunk-30
-fig:unnamed-chunk-35
-fig:unnamed-chunk-38
-fig:unnamed-chunk-40
+fig:unnamed-chunk-32
+fig:unnamed-chunk-37
fig:unnamed-chunk-42
-fig:unnamed-chunk-44
+fig:unnamed-chunk-45
+fig:unnamed-chunk-47
fig:unnamed-chunk-49
fig:unnamed-chunk-51
-fig:unnamed-chunk-53
-fig:unnamed-chunk-55
-fig:unnamed-chunk-59
-fig:unnamed-chunk-61
-fig:unnamed-chunk-63
+fig:unnamed-chunk-56
+fig:unnamed-chunk-58
+fig:unnamed-chunk-60
+fig:unnamed-chunk-62
fig:unnamed-chunk-66
-fig:unnamed-chunk-69
-fig:unnamed-chunk-74
+fig:unnamed-chunk-68
+fig:unnamed-chunk-70
+fig:unnamed-chunk-73
fig:unnamed-chunk-76
-fig:unnamed-chunk-78
-fig:unnamed-chunk-80
-fig:unnamed-chunk-82
-fig:unnamed-chunk-84
+fig:unnamed-chunk-81
+fig:unnamed-chunk-83
+fig:unnamed-chunk-85
fig:unnamed-chunk-87
+fig:unnamed-chunk-89
+fig:unnamed-chunk-91
fig:unnamed-chunk-94
-fig:unnamed-chunk-96
-fig:unnamed-chunk-98
fig:unnamed-chunk-101
fig:unnamed-chunk-103
fig:unnamed-chunk-105
+fig:unnamed-chunk-108
+fig:unnamed-chunk-110
fig:unnamed-chunk-112
-fig:unnamed-chunk-114
-fig:unnamed-chunk-118
-fig:unnamed-chunk-120
-fig:unnamed-chunk-123
-fig:unnamed-chunk-128
-fig:unnamed-chunk-131
-fig:unnamed-chunk-134
-fig:unnamed-chunk-137
+fig:unnamed-chunk-119
+fig:unnamed-chunk-121
+fig:unnamed-chunk-125
+fig:unnamed-chunk-127
+fig:unnamed-chunk-130
+fig:unnamed-chunk-135
+fig:unnamed-chunk-138
fig:unnamed-chunk-141
-fig:unnamed-chunk-143
-fig:unnamed-chunk-147
+fig:unnamed-chunk-144
+fig:unnamed-chunk-148
fig:unnamed-chunk-150
-fig:unnamed-chunk-152
+fig:unnamed-chunk-154
+fig:unnamed-chunk-157
fig:unnamed-chunk-159
-fig:unnamed-chunk-161
-fig:unnamed-chunk-163
-fig:unnamed-chunk-165
-fig:unnamed-chunk-167
-fig:unnamed-chunk-169
-fig:unnamed-chunk-171
+fig:unnamed-chunk-166
+fig:unnamed-chunk-168
+fig:unnamed-chunk-170
+fig:unnamed-chunk-172
fig:unnamed-chunk-174
-fig:unnamed-chunk-175
fig:unnamed-chunk-176
-fig:unnamed-chunk-185
-fig:unnamed-chunk-191
-fig:unnamed-chunk-194
+fig:unnamed-chunk-178
+fig:unnamed-chunk-181
+fig:unnamed-chunk-182
+fig:unnamed-chunk-183
+fig:unnamed-chunk-192
+fig:unnamed-chunk-198
fig:unnamed-chunk-201
-fig:unnamed-chunk-204
-fig:unnamed-chunk-206
-fig:unnamed-chunk-210
-fig:unnamed-chunk-212
-fig:unnamed-chunk-215
+fig:unnamed-chunk-208
+fig:unnamed-chunk-211
+fig:unnamed-chunk-213
fig:unnamed-chunk-217
fig:unnamed-chunk-219
-fig:unnamed-chunk-221
+fig:unnamed-chunk-222
fig:unnamed-chunk-224
-fig:unnamed-chunk-225
-fig:unnamed-chunk-227
-fig:unnamed-chunk-229
-fig:unnamed-chunk-230
+fig:unnamed-chunk-226
+fig:unnamed-chunk-228
+fig:unnamed-chunk-231
fig:unnamed-chunk-232
fig:unnamed-chunk-234
fig:unnamed-chunk-236
-fig:unnamed-chunk-275
-fig:unnamed-chunk-279
-fig:unnamed-chunk-281
-fig:unnamed-chunk-283
+fig:unnamed-chunk-237
+fig:unnamed-chunk-239
+fig:unnamed-chunk-241
+fig:unnamed-chunk-243
+fig:unnamed-chunk-282
fig:unnamed-chunk-286
fig:unnamed-chunk-288
fig:unnamed-chunk-290
-fig:unnamed-chunk-292
-fig:unnamed-chunk-294
+fig:unnamed-chunk-293
+fig:unnamed-chunk-295
fig:unnamed-chunk-297
fig:unnamed-chunk-299
fig:unnamed-chunk-301
-fig:unnamed-chunk-303
-fig:unnamed-chunk-305
-fig:unnamed-chunk-365
-fig:unnamed-chunk-366
+fig:unnamed-chunk-304
+fig:unnamed-chunk-306
+fig:unnamed-chunk-308
+fig:unnamed-chunk-310
+fig:unnamed-chunk-312
fig:unnamed-chunk-372
+fig:unnamed-chunk-373
fig:unnamed-chunk-379
-fig:unnamed-chunk-381
-fig:unnamed-chunk-387
-fig:unnamed-chunk-389
-fig:unnamed-chunk-395
-fig:unnamed-chunk-397
-fig:unnamed-chunk-462
-fig:unnamed-chunk-465
+fig:unnamed-chunk-386
+fig:unnamed-chunk-388
+fig:unnamed-chunk-394
+fig:unnamed-chunk-396
+fig:unnamed-chunk-402
+fig:unnamed-chunk-404
fig:unnamed-chunk-469
-fig:unnamed-chunk-470
fig:unnamed-chunk-472
-fig:unnamed-chunk-474
fig:unnamed-chunk-476
-fig:unnamed-chunk-478
+fig:unnamed-chunk-477
fig:unnamed-chunk-479
fig:unnamed-chunk-481
+fig:unnamed-chunk-483
fig:unnamed-chunk-485
+fig:unnamed-chunk-486
fig:unnamed-chunk-488
-fig:unnamed-chunk-490
fig:unnamed-chunk-492
+fig:unnamed-chunk-495
+fig:unnamed-chunk-497
+fig:unnamed-chunk-499
giotto-suite-workshop-2024
instructors
topics-and-schedule
@@ -153,6 +154,10 @@ presentation-1
ecosystem
installation-python-environment
giotto-installation-1
+system-prerequisites
+installation-of-released-version
+installation-of-dev-branch-giotto-packages
+common-install-issues
python-environment
default-installation
custom-installs
diff --git a/search_index.json b/search_index.json
index 2736898..72ab7c3 100644
--- a/search_index.json
+++ b/search_index.json
@@ -1 +1 @@
-[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: Ensure your system To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak. pak::pak("drieslab/Giotto") 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
+[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: 4.3.1.1 System prerequisites for windows, Rtools needs to be installed a major dependency terra needs GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux 4.3.1.2 Installation of released version To install the currently released version of Giotto in a single step: pak::pak("drieslab/Giotto") This should automatically install all the Giotto dependencies and other Giotto module packages. 4.3.1.3 Installation of dev branch Giotto packages You can install dev branch versions by using devtools::install_github() Core module dev branchs: “drieslab/Giotto@suite_dev” “drieslab/GiottoVisuals@dev” “drieslab/GiottoClass@dev” “drieslab/GiottoUtils@dev” devtools::install_github("drieslab/GiottoClass@dev") pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages. 4.3.1.4 Common install issues If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work. devtools::install_version("Matrix", version = "1.6-5") If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it. install.packages("irlba", type = "source") ## # Downloading packages ------------------------------------------------------- ## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s] ## Successfully downloaded 1 package in 1.2 seconds. ## ## The following package(s) will be installed: ## - irlba [2.3.5.1] ## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu". ## ## # Installing packages -------------------------------------------------------- ## - Installing irlba ... OK [built from source and cached in 5.7s] ## Successfully installed 1 package in 5.7 seconds. 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are: python_path - when set, will activate a python environment save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render. save_plot - whether to save plots to the save_dir return_plot - whether to return the plot objects. When FALSE, only NULL is returned show_plot - whether to show the plot in the viewer These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function. library(Giotto) save_dir <- "results/01_session2/" # this call will also intialize the python env instrs <- createGiottoInstructions( save_dir = save_dir, # working directory is the default show_plot = FALSE, save_plot = TRUE, return_plot = FALSE, python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default ) force(instrs) Giotto object creation functions all have an instructions param for passing in instructions objects. giotto objects will also respond to the instructions() generic. test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object # example plot g <- GiottoData::loadGiottoMini("visium") instructions(g) <- instrs instructions(g, "show_plot") # instructions say not to plot to viewer spatPlot2D(g, show_image = TRUE, image_name = "image") # instead it will directly write to the results folder As an example, you can also set individual instructions instructions(g, "show_plot") <- TRUE spatPlot2D(g, show_image = TRUE, image_name = "image") Figure 4.2: example image output "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
diff --git a/spatial-multi-modal-analysis.html b/spatial-multi-modal-analysis.html
index 5bdab74..c78f952 100644
--- a/spatial-multi-modal-analysis.html
+++ b/spatial-multi-modal-analysis.html
@@ -563,48 +563,48 @@ 13.2.3 Spatial classes:# load in data
-library(Giotto)
-
-g <- GiottoData::loadGiottoMini("vizgen")
-
-activeSpatUnit(g) <- "aggregate"
-
-gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE)
-
-gimg <- getGiottoImage(g)
+
13.3 Examples of the simple transforms with a giottoPolygon
-rain <- rainbow(nrow(gpoly))
-
-line_width <- 0.3
-
-# par to setup the grid plotting layout
-p <- par(no.readonly = TRUE)
-par(mfrow=c(3,3))
-gpoly |>
- plot(main = "no transform", col = rain, lwd = line_width)
-
-gpoly |> spatShift(dx = 1000) |>
- plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
-
-gpoly |> spin(45) |>
- plot(main = "spin(45)", col = rain, lwd = line_width)
-
-gpoly |> rescale(fx = 10, fy = 6) |>
- plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
-
-gpoly |> flip(direction = "vertical") |>
- plot(main = "flip()", col = rain, lwd = line_width)
-
-gpoly |> t() |>
- plot(main = "t()", col = rain, lwd = line_width)
-
-gpoly |> shear(fx = 0.5) |>
- plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
-par(p)
+rain <- rainbow(nrow(gpoly))
+
+line_width <- 0.3
+
+# par to setup the grid plotting layout
+p <- par(no.readonly = TRUE)
+par(mfrow=c(3,3))
+gpoly |>
+ plot(main = "no transform", col = rain, lwd = line_width)
+
+gpoly |> spatShift(dx = 1000) |>
+ plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
+
+gpoly |> spin(45) |>
+ plot(main = "spin(45)", col = rain, lwd = line_width)
+
+gpoly |> rescale(fx = 10, fy = 6) |>
+ plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
+
+gpoly |> flip(direction = "vertical") |>
+ plot(main = "flip()", col = rain, lwd = line_width)
+
+gpoly |> t() |>
+ plot(main = "t()", col = rain, lwd = line_width)
+
+gpoly |> shear(fx = 0.5) |>
+ plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
+par(p)
@@ -617,20 +617,20 @@ 13.4 Affine transforms# create affine2d
-aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
+# create affine2d
+aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set.
-
-# append several simple transforms
-aff <- aff |>
- spatShift(dx = 1000) |>
- spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- flip(direction = "vertical") |>
- t() |>
- shear(fx = 0.5)
-force(aff)
+
+# append several simple transforms
+aff <- aff |>
+ spatShift(dx = 1000) |>
+ spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ flip(direction = "vertical") |>
+ t() |>
+ shear(fx = 0.5)
+force(aff)
<affine2d>
anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax)
rotate : -0.785398163397448 (rad)
@@ -639,35 +639,35 @@ 13.4 Affine transformsplot(aff)
+
We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests.
-
+
13.5 Image transforms
Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting.
spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image.
-# affine transform of image info matches with polygon info
-gimg |> affine(aff) |> plot()
-gpoly |> affine(aff) |>
- plot(add = TRUE,
- border = "cyan",
- lwd = 0.3)
+# affine transform of image info matches with polygon info
+gimg |> affine(aff) |> plot()
+gpoly |> affine(aff) |>
+ plot(add = TRUE,
+ border = "cyan",
+ lwd = 0.3)
-# affine of the giotto object
-g |> affine(aff) |>
- spatInSituPlotPoints(
- show_image = TRUE,
- feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
- feats_color_code = rainbow(4),
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- point_size = 0.1,
- use_overlap = FALSE
- )
+# affine of the giotto object
+g |> affine(aff) |>
+ spatInSituPlotPoints(
+ show_image = TRUE,
+ feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
+ feats_color_code = rainbow(4),
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ point_size = 0.1,
+ use_overlap = FALSE
+ )
Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used.
@@ -687,256 +687,256 @@ 13.6.1 Example dataset: Xenium Br
– cooresponding centroid locations
The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance.
-library(Giotto)
-instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
- save_plot = TRUE,
- show_plot = TRUE)
-options(timeout = 999999)
-download_dir <-file.path(getwd(),'/data/03_session2/')
-destfile <- file.path(download_dir,'Multimodal_registration.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
-Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
-Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
+library(Giotto)
+instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
+ save_plot = TRUE,
+ show_plot = TRUE)
+options(timeout = 999999)
+download_dir <-file.path(getwd(),'/data/03_session2/')
+destfile <- file.path(download_dir,'Multimodal_registration.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
+Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
+Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
13.6.2 Target Coordinate system
Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system.
-xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
-xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
-xen_cell_pl
+xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
+xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+xen_cell_pl
13.6.3 Visium to register
Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration
-G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
- gene_column_index = 2,
- png_name = 'tissue_hires_image.png',
- instructions = NULL)
-# In the meantime, calculate statistics for easier plot showing
-G_visium <- normalizeGiotto(G_visium)
-G_visium <- addStatistics(G_visium)
-V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
-V_origin
+G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
+ gene_column_index = 2,
+ png_name = 'tissue_hires_image.png',
+ instructions = NULL)
+# In the meantime, calculate statistics for easier plot showing
+G_visium <- normalizeGiotto(G_visium)
+G_visium <- addStatistics(G_visium)
+V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
+V_origin
The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform.
-# create affine2d
-aff <- affine(diag(c(1,1)))
-aff <- aff |>
- spin(90) |>
- flip(direction = "horizontal")
-force(aff)
-
-# Apply the transform
-V_tansformed <- affine(G_visium,aff)
-spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
-spatplot_to_register
+# create affine2d
+aff <- affine(diag(c(1,1)))
+aff <- aff |>
+ spin(90) |>
+ flip(direction = "horizontal")
+force(aff)
+
+# Apply the transform
+V_tansformed <- affine(G_visium,aff)
+spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
+spatplot_to_register
Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images.
Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app.
-
+
Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result.
-landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
-affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
-
-V_final <- affine(G_visium,affine_mtx %*% aff@affine)
-
-spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
-spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
+affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
+
+V_final <- affine(G_visium,affine_mtx %*% aff@affine)
+
+spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
+spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.3.1 Create Pseudo Visium dataset for comparison
Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance.
-V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
-spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
-spatPlot2D(V_final,
- show_network = T,
- network_color = 'blue',
- point_size = 1)
-center_to_center <- min(spat_network$distance)
-radius <- center_to_center*55/200
+V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
+spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
+spatPlot2D(V_final,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1)
+center_to_center <- min(spat_network$distance)
+radius <- center_to_center*55/200
Now we get the Pseudo Visium polygons
-Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
-stamp_dt <- circleVertices(radius = radius, npoints = 100)
-pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
-pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
-plot(pseudo_visium_poly)
+Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
+stamp_dt <- circleVertices(radius = radius, npoints = 100)
+pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
+pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
+plot(pseudo_visium_poly)
Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points
-xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
-gpoints <- createGiottoPoints(xen_transcripts)
-Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
- gpolygons = list('visium' = pseudo_visium_poly))
+xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
+gpoints <- createGiottoPoints(xen_transcripts)
+Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
+ gpolygons = list('visium' = pseudo_visium_poly))
Get gene expression information by overlapping polygon to points
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'visium')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "visium",
- feat_info = "rna",
- aggr_function = "sum")
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'visium')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "visium",
+ feat_info = "rna",
+ aggr_function = "sum")
Manipulate the expression for plotting
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'visium',
- name = 'plot')
-
-spatFeatPlot2D(Xen_obj,
- point_size = 3.5,
- expression_values = 'plot',
- show_image = F,
- feats = 'ERBB2')
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ name = 'plot')
+
+spatFeatPlot2D(Xen_obj,
+ point_size = 3.5,
+ expression_values = 'plot',
+ show_image = F,
+ feats = 'ERBB2')
Subset the registered Visium and plot same gene
-#get the extent of giotto points, xmin, xmax, ymin, ymax
-subset_extent <- ext(gpoints@spatVector)
-sub_visium <- subsetGiottoLocs(V_final,
- x_min = subset_extent[1],
- x_max = subset_extent[2],
- y_min = subset_extent[3],
- y_max = subset_extent[4])
-spatFeatPlot2D(sub_visium,
- point_size = 2,
- expression_values = 'scaled',
- show_image = F,
- feats = 'ERBB2')
+#get the extent of giotto points, xmin, xmax, ymin, ymax
+subset_extent <- ext(gpoints@spatVector)
+sub_visium <- subsetGiottoLocs(V_final,
+ x_min = subset_extent[1],
+ x_max = subset_extent[2],
+ y_min = subset_extent[3],
+ y_max = subset_extent[4])
+spatFeatPlot2D(sub_visium,
+ point_size = 2,
+ expression_values = 'scaled',
+ show_image = F,
+ feats = 'ERBB2')
13.6.4 Register post-Xenium H&E and IF image
For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics.
-HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
- imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
- micron = 0.2125)
-plot(HE_xen)
+HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
+ imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
+ micron = 0.2125)
+plot(HE_xen)
The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use.
-HE_xen <- HE_xen |> flip(direction = "vertical")
-gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
-plot(gimg_rast)
-#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
+HE_xen <- HE_xen |> flip(direction = "vertical")
+gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
+plot(gimg_rast)
+#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots
-gg <- ggplot2::ggplot()
-pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
-pl + ggplot2::geom_smooth() +
- ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+gg <- ggplot2::ggplot()
+pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
+pl + ggplot2::geom_smooth() +
+ ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.4.1 Add registered image information and compare RNA vs protein expression
With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF
-CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
-HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
-Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
- largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
+CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
+HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
+Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
+ largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
Get the cell polygons, as Xenium and IF are both subcellular resolution
-cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
-colnames(cellpoly_dt) <- c('poly_ID','x','y')
-cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
-Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
+cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
+colnames(cellpoly_dt) <- c('poly_ID','x','y')
+cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
+Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
Compute the gene expression matrix by overlay the cell polygons and giotto points.
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'cell')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "cell",
- feat_info = "rna",
- aggr_function = "sum")
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'rna',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'ERBB2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'cell')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "cell",
+ feat_info = "rna",
+ aggr_function = "sum")
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'rna',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'ERBB2',
+ point_size = 0.05)
Now we overlay the HER2 expression from the raster image with the cell polygons.
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'cell',
- image_names = c('HER2','CD20'))
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'HER2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'cell',
+ image_names = c('HER2','CD20'))
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'HER2',
+ point_size = 0.05)
We can also overlay the protein expression to Visium spots
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'visium',
- image_names = c('HER2','CD20'))
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "visium",
- feat_info = "protein",
- aggr_function = "sum")
-
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'visium',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'visium',
- feats = 'HER2',
- point_size = 2)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'visium',
+ image_names = c('HER2','CD20'))
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "visium",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'visium',
+ feats = 'HER2',
+ point_size = 2)
@@ -944,29 +944,29 @@ 13.6.4.1 Add registered image inf
13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation
Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially.
Here, we provide an example of two compressed images to show the automatic alignment pipeline.
-
+
-IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
-terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
+IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
+terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform.
-estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
- flip_horizontal = T,
- use_single_channel = T,
- single_channel_number = 3),
- y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
- invert = T),
- plot_match = T,
- max_ratio = 0.5,estimate_fun = 'Projective')
+estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
+ flip_horizontal = T,
+ use_single_channel = T,
+ single_channel_number = 3),
+ y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
+ invert = T),
+ plot_match = T,
+ max_ratio = 0.5,estimate_fun = 'Projective')
Use the estimation, we can quickly visualize the transformation
-mtx <- as.matrix(estimation$params)
-transformed <- affine(IF, mtx)
-To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
-plot(HE)
-plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
-SIFT_registration_overlay
+mtx <- as.matrix(estimation$params)
+transformed <- affine(IF, mtx)
+To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
+plot(HE)
+plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
+SIFT_registration_overlay
diff --git a/spatial-proteomics-multiplexed-immunofluorescence.html b/spatial-proteomics-multiplexed-immunofluorescence.html
index 684d4e8..f550962 100644
--- a/spatial-proteomics-multiplexed-immunofluorescence.html
+++ b/spatial-proteomics-multiplexed-immunofluorescence.html
@@ -518,33 +518,33 @@ 11 Spatial proteomics: Multiplexe
Junxiang Xu
August 6th 2024
Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so.
-#Install the Giotto Environment with Cellpose, note that we only need to do it once
-reticulate::conda_create(envname = "giotto_cellpose",
- python_version = 3.8)
-#.re.restartR()
-reticulate::use_condaenv('giotto_cellpose')
-reticulate::py_install(
- pip = TRUE,
- envname = 'giotto_cellpose',
- packages = c(
- "pandas",
- "networkx",
- "python-igraph",
- "leidenalg",
- "scikit-learn",
- "cellpose",
- "smfishhmrf",
- 'tifffile',
- 'scikit-image'
- )
-)
-#.rs.restartR()
+#Install the Giotto Environment with Cellpose, note that we only need to do it once
+reticulate::conda_create(envname = "giotto_cellpose",
+ python_version = 3.8)
+#.re.restartR()
+reticulate::use_condaenv('giotto_cellpose')
+reticulate::py_install(
+ pip = TRUE,
+ envname = 'giotto_cellpose',
+ packages = c(
+ "pandas",
+ "networkx",
+ "python-igraph",
+ "leidenalg",
+ "scikit-learn",
+ "cellpose",
+ "smfishhmrf",
+ 'tifffile',
+ 'scikit-image'
+ )
+)
+#.rs.restartR()
Now, activate the Giotto python environment.
-#.rs.restartR()
-# Activate the Giotto python environment of your choice
-GiottoClass::set_giotto_python_path('giotto_cellpose')
-# Check if cellpose was successfully installed
-GiottoUtils::package_check('cellpose',repository = 'pip')
+#.rs.restartR()
+# Activate the Giotto python environment of your choice
+GiottoClass::set_giotto_python_path('giotto_cellpose')
+# Check if cellpose was successfully installed
+GiottoUtils::package_check('cellpose',repository = 'pip')
11.1 Spatial Proteomics Technologies
This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix.
@@ -557,58 +557,58 @@
11.2 Raw data type coming out of
11.2.1 Use ome.tiff as an example output data to begin with
OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore.
-library(Giotto)
-instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
- save_plot = TRUE,
- show_plot = TRUE,
- python_path = 'giotto_cellpose')
-options(timeout = 9999999)
-download_dir =file.path(getwd(),'/data/02_session4/')
-destfile = file.path(download_dir,'Lunaphore.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
-list.files(paste0(download_dir,'/Lunaphore'))
+library(Giotto)
+instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ python_path = 'giotto_cellpose')
+options(timeout = 9999999)
+download_dir =file.path(getwd(),'/data/02_session4/')
+destfile = file.path(download_dir,'Lunaphore.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
+list.files(paste0(download_dir,'/Lunaphore'))
We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML.
-img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
-img_meta
+img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
+img_meta
However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it.
-##Get channel information
-Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
-node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
-channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
-channel_df
+##Get channel information
+Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
+node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
+channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
+channel_df
11.2.2 Use single channel images as an example output data to begin with
Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like.
-#Create multichannel raster and extract each single channels
-Luna_terra = terra::rast(Luna)
-names(Luna_terra) <- channel_df$Name
-gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
-plot(gimg_DAPI)
+#Create multichannel raster and extract each single channels
+Luna_terra = terra::rast(Luna)
+names(Luna_terra) <- channel_df$Name
+gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
+plot(gimg_DAPI)
Extract and save the raster image for future use.
-single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
-if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
-for (i in 1:nrow(channel_df)){
- single_channel <- terra::subset(Luna_terra, i)
- terra::writeRaster(single_channel,
- filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
- overwrite=T)
-}
+single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
+if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
+for (i in 1:nrow(channel_df)){
+ single_channel <- terra::subset(Luna_terra, i)
+ terra::writeRaster(single_channel,
+ filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
+ overwrite=T)
+}
Create a list of GiottoLargeImages using single channel rasters.
-file_names = list.files(single_channel_dir,full.names = TRUE)
-image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
-gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
- names = image_names,
- negative_y = F,
- flip_vertical = T)
-names(gimg_list) <- image_names
-plot(gimg_list[['Vimentin']])
+file_names = list.files(single_channel_dir,full.names = TRUE)
+image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
+gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
+ names = image_names,
+ negative_y = F,
+ flip_vertical = T)
+names(gimg_list) <- image_names
+plot(gimg_list[['Vimentin']])
@@ -618,34 +618,34 @@ 11.3 Cell Segmentation file to ge
11.3.1 Using segmentation output file from DeepCell(mesmer) as an example.
We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function.
-gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
-plot(gpoly_mesmer)
+gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
+plot(gpoly_mesmer)
We can also zoom in to check how does the segmentation look.
-zoom <- c(2000,2500,2000,2500)
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+zoom <- c(2000,2500,2000,2500)
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
11.3.2 Using Giotto wrapper of Cellpose to perform segmentation
-gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
-writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
-#Create a giotto image to evaluate segmentation
-gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
+gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
+writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
+#Create a giotto image to evaluate segmentation
+gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website
-doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
- mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- channel_1 = 0,
- channel_2 = 0,
- model_name = 'cyto3',
- batch_size=12)
-cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- shift_horizontal_step = F,
- shift_vertical_step = F,
- flip_vertical = T)
-plot(gimg_for_cellpose)
-plot(cpoly, add = T, border = 'red')
+doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
+ mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ channel_1 = 0,
+ channel_2 = 0,
+ model_name = 'cyto3',
+ batch_size=12)
+cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ shift_horizontal_step = F,
+ shift_vertical_step = F,
+ flip_vertical = T)
+plot(gimg_for_cellpose)
+plot(cpoly, add = T, border = 'red')
@@ -654,133 +654,133 @@ 11.4 Create a Giotto Object using
You will need to have
- list of giotto images
- giottoPolygon created from segmentation
-Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
- images = gimg_list,
- instructions = instrs)
-Lunaphore_giotto
+Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
+ images = gimg_list,
+ instructions = instrs)
+Lunaphore_giotto
11.4.1 Overlap to matrix
calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with
-Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
- spatial_info = 'cell',
- image_names = names(gimg_list))
-
-Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
- type ='intensity',
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-showGiottoExpression(Lunaphore_giotto)
+Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
+ spatial_info = 'cell',
+ image_names = names(gimg_list))
+
+Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
+ type ='intensity',
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+showGiottoExpression(Lunaphore_giotto)
11.4.2 Manipulate Expression information
For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile.
Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object
-expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
-filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
-Lunaphore_giotto <- setExpression(Lunaphore_giotto,
- feat_type = 'protein',
- x = createExprObj(filtered_expr_mtx),
- name = 'raw')
-showGiottoExpression(Lunaphore_giotto)
+expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
+filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
+Lunaphore_giotto <- setExpression(Lunaphore_giotto,
+ feat_type = 'protein',
+ x = createExprObj(filtered_expr_mtx),
+ name = 'raw')
+showGiottoExpression(Lunaphore_giotto)
11.4.3 Rescale polygons
rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis
-Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
- poly_info = 'cell',
- name = 'smallcell',
- fx = 0.7, fy = 0.7, calculate_centroids = T)
-smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
-plot(smallpoly,add = TRUE, border = "red",ext = zoom)
+Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
+ poly_info = 'cell',
+ name = 'smallcell',
+ fx = 0.7, fy = 0.7, calculate_centroids = T)
+smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+plot(smallpoly,add = TRUE, border = "red",ext = zoom)
11.4.4 Perform clustering and differential expression
The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering
-Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
-Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
-
-Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
- scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
- set_seed = TRUE)
-screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
+Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
+Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
+
+Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
+ scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
+ set_seed = TRUE)
+screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check.
-Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
- dimensions_to_use = 1:5,
- set_seed = TRUE)
-Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
-Lunaphore_giotto = doHclust(Lunaphore_giotto,
- k = 8,
- dim_reduction_to_use = 'cells',
- spat_unit = 'cell',
- feat_type = 'protein')
-
-spatInSituPlotPoints(gobject = Lunaphore_giotto,
- spat_unit = "cell",
- polygon_feat_type = "cell",
- show_polygon = T,
- feat_type = "protein",
- feats = NULL,
- polygon_fill = 'hclust',
- polygon_fill_as_factor = TRUE,
- polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
- polygon_color = 'black',
- background_color = "white")
+Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
+ dimensions_to_use = 1:5,
+ set_seed = TRUE)
+Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
+Lunaphore_giotto = doHclust(Lunaphore_giotto,
+ k = 8,
+ dim_reduction_to_use = 'cells',
+ spat_unit = 'cell',
+ feat_type = 'protein')
+
+spatInSituPlotPoints(gobject = Lunaphore_giotto,
+ spat_unit = "cell",
+ polygon_feat_type = "cell",
+ show_polygon = T,
+ feat_type = "protein",
+ feats = NULL,
+ polygon_fill = 'hclust',
+ polygon_fill_as_factor = TRUE,
+ polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
+ polygon_color = 'black',
+ background_color = "white")
Then we can check the heatmap of protein expression and determine the first round of cluster annotation
-cluster_column = 'hclust'
-plotMetaDataHeatmap(Lunaphore_giotto,
- spat_unit = 'cell',feat_type = 'protein',
- expression_values = "raw",
- metadata_cols = c(cluster_column),
- selected_feats = names(gimg_list),
- y_text_size = 8,
- show_values = 'zscores_rescaled')
+cluster_column = 'hclust'
+plotMetaDataHeatmap(Lunaphore_giotto,
+ spat_unit = 'cell',feat_type = 'protein',
+ expression_values = "raw",
+ metadata_cols = c(cluster_column),
+ selected_feats = names(gimg_list),
+ y_text_size = 8,
+ show_values = 'zscores_rescaled')
11.4.5 Give the cluster an annotation based on expression values
-
+
11.4.6 spatial network
This is to create a cellular neighborhood based on nearest neighbor of physical distance
-Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
-
-spatPlot2D(Lunaphore_giotto,
- show_network = T,
- network_color = 'blue',
- point_size = 1.5,
- cell_color = 'hclust')
+Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
+
+spatPlot2D(Lunaphore_giotto,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1.5,
+ cell_color = 'hclust')
11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions
This is using cellProximityEnrichment() to statistically identify cell type interactions
-cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
- cluster_column = 'cell_types',
- spatial_network_name = 'Delaunay_network',
- adjust_method = 'fdr',
- number_of_simulations = 2000)
-## barplot
-cellProximityBarplot(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- min_orig_ints = 5, min_sim_ints = 5)
+cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
+ cluster_column = 'cell_types',
+ spatial_network_name = 'Delaunay_network',
+ adjust_method = 'fdr',
+ number_of_simulations = 2000)
+## barplot
+cellProximityBarplot(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ min_orig_ints = 5, min_sim_ints = 5)
-## network
-cellProximityNetwork(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- remove_self_edges = T,
- only_show_enrichment_edges = F)
+## network
+cellProximityNetwork(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ remove_self_edges = T,
+ only_show_enrichment_edges = F)
diff --git a/visium-hd.html b/visium-hd.html
index 23aa210..f24cdd4 100644
--- a/visium-hd.html
+++ b/visium-hd.html
@@ -525,7 +525,7 @@ 9.1 Objective9.2 Background
9.2.1 Visium HD Technology
-
+
Figure 9.1: Overview of Visium HD. Source: 10X Genomics
@@ -536,7 +536,7 @@
9.2.1 Visium HD Technology
9.2.2 Colorectal Cancer Sample
-
+
Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics
@@ -550,7 +550,7 @@
9.2.2 Colorectal Cancer Sample9.3 Data Ingestion
9.3.1 Visium HD output data format
-
+
Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline.
@@ -560,19 +560,19 @@
9.3.1 Visium HD output data forma
9.4 Tiling and aggregation
The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells.
-
+
9.5 Scalability and projection functions
diff --git a/visium-part-i.html b/visium-part-i.html
index 5aa1658..c4746ee 100644
--- a/visium-part-i.html
+++ b/visium-part-i.html
@@ -520,7 +520,7 @@ 7 Visium Part I
7.1 The Visium technology
Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression.
-
+
Figure 7.1: Visum workflow. Source: 10X Genomics
@@ -536,73 +536,73 @@
7.2 Introduction to the spatial d
7.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/01_session5/")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
+dir.create("data/01_session5/")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”.
-
+
7.4 Create the Giotto object
createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/01_session5/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-## Provide the path to the visium folder
-data_path <- "data/01_session5/"
-
-## Create object directly from the visium folder
-visium_brain <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/01_session5/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+## Provide the path to the visium folder
+data_path <- "data/01_session5/"
+
+## Create object directly from the visium folder
+visium_brain <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
7.5 Subset on spots that were covered by tissue
Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area.
-spatPlot2D(
- gobject = visium_brain,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE)
-
+spatPlot2D(
+ gobject = visium_brain,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE)
+
Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area.
-
+
7.6 Quality control
@@ -610,43 +610,43 @@ 7.6 Quality controlvisium_brain_statistics <- addStatistics(gobject = visium_brain,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = visium_brain_statistics,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain_statistics <- addStatistics(gobject = visium_brain,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = visium_brain_statistics,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.3: Spatial distribution of features per spot.
filterDistributions() creates a histogram to show the distribution of features per spot across the sample.
-
-
+
+
Figure 7.4: Distribution of features per spot.
When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample.
-
-
+
+
Figure 7.5: Distribution of cells with different features per spot.
filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data:
-filterCombinations(gobject = visium_brain_statistics,
- expression_thresholds = c(1, 2, 3),
- feat_det_in_min_cells = c(50, 100, 200),
- min_det_feats_per_cell = c(500, 1000, 1500))
-
+filterCombinations(gobject = visium_brain_statistics,
+ expression_thresholds = c(1, 2, 3),
+ feat_det_in_min_cells = c(50, 100, 200),
+ min_det_feats_per_cell = c(500, 1000, 1500))
+
Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations.
@@ -656,34 +656,34 @@
7.6 Quality control
7.7 Filtering
Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample.
-visium_brain <- filterGiotto(
- gobject = visium_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE
-)
-Feature type: rna
-Number of cells removed: 4 out of 2702
-Number of feats removed: 7311 out of 22125
+
+
7.8 Normalization
Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one.
-visium_brain <- normalizeGiotto(
- gobject = visium_brain,
- scalefactor = 6000,
- verbose = TRUE
-)
+visium_brain <- normalizeGiotto(
+ gobject = visium_brain,
+ scalefactor = 6000,
+ verbose = TRUE
+)
Calculate the normalized number of features per spot and save the statistics in the metadata table.
-visium_brain <- addStatistics(gobject = visium_brain)
-
-## visualize
-spatPlot2D(gobject = visium_brain,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain <- addStatistics(gobject = visium_brain)
+
+## visualize
+spatPlot2D(gobject = visium_brain,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.7: Spatial distribution of the number of features per spot.
@@ -698,11 +698,11 @@
7.9.1 Highly Variable Features:
loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_loess",
- save_plot = TRUE,
- default_save_name = "HVFplot_loess")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_loess",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_loess")
+
Figure 7.8: Covariance of HVFs using the loess method.
@@ -711,11 +711,11 @@
7.9.1 Highly Variable Features:
pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")
+
Figure 7.9: Variance of HVFs using the pearson residuals method.
@@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
17.4.4 Adding polygons to giotto object
After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject.
-v_brain <- addGiottoPolygons(v_brain,
- gpolygons = list('stardist_cell' = stardist_cell_gpoly))
-
-print(v_brain)
+
+
17.4.4 Adding polygons to giotto
v_brain <- addGiottoPolygons(v_brain,
+ gpolygons = list('stardist_cell' = stardist_cell_gpoly))
+
+print(v_brain)Figure 17.6: Giotto object after to adding subcellular polygons. @@ -711,9 +711,9 @@
17.4.4 Adding polygons to giotto
17.4.5 Check polygon information
We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots.
-
-
+
+
Figure 17.7: Polygon information for stardist_cell.
@@ -727,23 +727,23 @@
17.5 Performing kriging17.5.1 Interpolating features
Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be
analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required.
-future::plan(future::multisession()) # comment out for single threading
-subcell_v_brain <- interpolateFeature(v_brain,
- spat_unit = 'cell',
- feat_type = 'rna',
- ext = ext(v_brain),
- feats = ext_spatial_features,
- overwrite = T)
-
-print(subcell_v_brain)
-
+future::plan(future::multisession()) # comment out for single threading
+subcell_v_brain <- interpolateFeature(v_brain,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ ext = ext(v_brain),
+ feats = ext_spatial_features,
+ overwrite = T)
+
+print(subcell_v_brain)
+
Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right).
For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1.
-
+
Figure 17.9: Raster of gene expression interpolation for Pantr1
@@ -754,18 +754,18 @@
17.5.1 Interpolating features17.5.2 Expression overlap
The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier.
This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run.
-subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain,
- name_overlap = "rna",
- spatial_info = "stardist_cell",
- image_names = ext_spatial_features)
-
-subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain,
- poly_info = "stardist_cell",
- feat_info = "rna",
- aggr_function = "sum",
- type='intensity')
+subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain,
+ name_overlap = "rna",
+ spatial_info = "stardist_cell",
+ image_names = ext_spatial_features)
+
+subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain,
+ poly_info = "stardist_cell",
+ feat_info = "rna",
+ aggr_function = "sum",
+ type='intensity')
After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified.
-
+
Figure 17.10: Gene expression for cells based on interpolation.
@@ -776,70 +776,70 @@
17.5.2 Expression overlap
17.6 Reading in larger dataset
For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here.
-
+
17.7 Analyzing interpolated features
17.7.1 Filter and normalization
Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis.
-subcell_v_brain <- filterGiotto(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- expression_values = "raw",
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-
-subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- scalefactor = 6000,
- verbose = T)
+subcell_v_brain <- filterGiotto(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ expression_values = "raw",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+
+subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ scalefactor = 6000,
+ verbose = T)
17.7.2 Visualizing gene expression from interpolated expression
Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data.
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Sparc',
- point_size = 2,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_sparc_vis'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'stardist_cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Sparc',
- point_size = 0.6,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_sparc'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'cell',
- gradient_style = 'sequential',
- feats = 'Pantr1',
- cell_color_gradient = 'Geyser',
- point_size = 2,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_pantr1_vis'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'stardist_cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Pantr1',
- point_size = 0.6,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_pantr1'))
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Sparc',
+ point_size = 2,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_sparc_vis'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Sparc',
+ point_size = 0.6,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_sparc'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'cell',
+ gradient_style = 'sequential',
+ feats = 'Pantr1',
+ cell_color_gradient = 'Geyser',
+ point_size = 2,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_pantr1_vis'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Pantr1',
+ point_size = 0.6,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_pantr1'))
Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer.
-
+
Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom).
@@ -848,37 +848,37 @@
17.7.2 Visualizing gene expressio
17.7.3 Run PCA
-
+
17.7.4 Clustering
-# UMAP
-subcell_v_brain <- runUMAP(subcell_v_brain,
- spat_unit = "stardist_cell",
- dimensions_to_use = 1:15,
- n_neighbors = 1000,
- min_dist = 0.001,
- spread = 1)
-
-# NN Network
-subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- dimensions_to_use = 1:10,
- feats_to_use = hv_feats,
- expression_values = 'normalized',
- k = 70)
-
-subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- resolution = 0.15,
- n_iterations = 100,
- partition_type = 'RBConfigurationVertexPartition')
-
-plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus')
-
+# UMAP
+subcell_v_brain <- runUMAP(subcell_v_brain,
+ spat_unit = "stardist_cell",
+ dimensions_to_use = 1:15,
+ n_neighbors = 1000,
+ min_dist = 0.001,
+ spread = 1)
+
+# NN Network
+subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ dimensions_to_use = 1:10,
+ feats_to_use = hv_feats,
+ expression_values = 'normalized',
+ k = 70)
+
+subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ resolution = 0.15,
+ n_iterations = 100,
+ partition_type = 'RBConfigurationVertexPartition')
+
+plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus')
+
Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering.
@@ -888,27 +888,27 @@
17.7.4 Clustering
17.7.5 Visualizing clustering
Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster.
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.5,
- point_shape = 'no_border',
- background_color = 'black',
- save_plot = F,
- show_legend = TRUE)
-
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat'))
-
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.5,
+ point_shape = 'no_border',
+ background_color = 'black',
+ save_plot = F,
+ show_legend = TRUE)
+
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat'))
+
Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
@@ -918,37 +918,37 @@
17.7.5 Visualizing clustering
17.7.6 Cropping objects
We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below.
-subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
- spat_unit = ":all:",
- x_min = 4000,
- x_max = 7000,
- y_min = -6500,
- y_max = -3500,
- z_max = NULL,
- z_min = NULL)
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 2,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_vis_spat_crop'))
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat_crop'))
-
+subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
+ spat_unit = ":all:",
+ x_min = 4000,
+ x_max = 7000,
+ y_min = -6500,
+ y_max = -3500,
+ z_max = NULL,
+ z_min = NULL)
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 2,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_vis_spat_crop'))
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat_crop'))
+
Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
diff --git a/interoperability-with-isolated-tools.html b/interoperability-with-isolated-tools.html
index ceb2d0a..8cca7e6 100644
--- a/interoperability-with-isolated-tools.html
+++ b/interoperability-with-isolated-tools.html
@@ -526,1521 +526,1521 @@
16.1.1 Dataset downloaddata_path <- "data"
-dir.create(data_path)
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
- destfile = file.path(data_path,"counts.h5ad"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
- destfile = file.path(data_path,"cell_labels.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
+data_path <- "data"
+dir.create(data_path)
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
+ destfile = file.path(data_path,"counts.h5ad"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
+ destfile = file.path(data_path,"cell_labels.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
16.1.2 Create the Giotto object
-library(Giotto)
-library(reticulate)
-
-## Set instructions
-results_folder <- "results"
-dir.create(results_folder)
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-
-## load data
-
-### meta data
-meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-colnames(meta_df)[[1]] <- "cell_ID"
-slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
-slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
-
-### counts matrix
-scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
-ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
-reticulate::source_python(ad2g_path)
-adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
-X <- extract_expression(adata)
-cID <- extract_cell_IDs(adata)
-fID <- extract_feat_IDs(adata)
-rownames(X) <- fID
-colnames(X) <- cID
-slice1_filter <- cID %in% slice1_meta_df$cell_ID
-slice2_filter <- cID %in% slice2_meta_df$cell_ID
-slice1_exp <- X[,slice1_filter]
-slice2_exp <- X[,slice2_filter]
-
-### cell segmentation to cell location
-segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_df <- rbind(segments_1_df, segments_2_df)
-loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
-loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
-loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
-centr.x <- apply(loc.use[,loc.x],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-centr.y <- apply(loc.use[,loc.y],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
-slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
-slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
-
-## giotto obj
-giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
- spatial_locs = slice1_spatial_locs_df,
- cell_metadata = slice1_meta_df)
-giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
- spatial_locs = slice2_spatial_locs_df,
- cell_metadata = slice2_meta_df)
-giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
- gobject_names = c('mouse2_slice229',
- 'mouse2_slice300'), # name for each samples
- join_method = 'z_stack')
-# We skip the processing process here to save time and use the given cell type
-# annotation directly
-
-ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
- cell_type = "subclass",
- output_path = data_path,
- spat_unit = "cell",
- feat_type = "rna",
- verbose = TRUE)
-
-# Cell_ID Sample x y Cell_Type
-# <char> <char> <dbl> <dbl> <char>
-# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
-# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
-# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
-# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
-# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
-# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
+library(Giotto)
+library(reticulate)
+
+## Set instructions
+results_folder <- "results"
+dir.create(results_folder)
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+
+## load data
+
+### meta data
+meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+colnames(meta_df)[[1]] <- "cell_ID"
+slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
+slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
+
+### counts matrix
+scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
+ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
+reticulate::source_python(ad2g_path)
+adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
+X <- extract_expression(adata)
+cID <- extract_cell_IDs(adata)
+fID <- extract_feat_IDs(adata)
+rownames(X) <- fID
+colnames(X) <- cID
+slice1_filter <- cID %in% slice1_meta_df$cell_ID
+slice2_filter <- cID %in% slice2_meta_df$cell_ID
+slice1_exp <- X[,slice1_filter]
+slice2_exp <- X[,slice2_filter]
+
+### cell segmentation to cell location
+segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_df <- rbind(segments_1_df, segments_2_df)
+loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
+loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
+loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
+centr.x <- apply(loc.use[,loc.x],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+centr.y <- apply(loc.use[,loc.y],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
+slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
+slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
+
+## giotto obj
+giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
+ spatial_locs = slice1_spatial_locs_df,
+ cell_metadata = slice1_meta_df)
+giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
+ spatial_locs = slice2_spatial_locs_df,
+ cell_metadata = slice2_meta_df)
+giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
+ gobject_names = c('mouse2_slice229',
+ 'mouse2_slice300'), # name for each samples
+ join_method = 'z_stack')
+# We skip the processing process here to save time and use the given cell type
+# annotation directly
+
+ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
+ cell_type = "subclass",
+ output_path = data_path,
+ spat_unit = "cell",
+ feat_type = "rna",
+ verbose = TRUE)
+
+# Cell_ID Sample x y Cell_Type
+# <char> <char> <dbl> <dbl> <char>
+# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
+# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
+# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
+# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
+# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
+# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
Spatial cell type distributions
-spatPlot2D(giotto_obj,
- group_by = "slice_id",
- cell_color = "subclass",
- point_size = 1,
- point_border_stroke = NA,
- legend_text = 6)
+spatPlot2D(giotto_obj,
+ group_by = "slice_id",
+ cell_color = "subclass",
+ point_size = 1,
+ point_border_stroke = NA,
+ legend_text = 6)
16.1.2.1 Installation ONTraC
You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node.
It will run for less than 10 minutes on this example dataset.
For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time.
-source ~/.bash_profile
-conda create -y -n ONTraC python=3.11
-conda activate ONTraC
-pip install git+https://github.com/gyuanlab/ONTraC.git@V2
-# Retrieving notices: ...working... done
-# Channels:
-# - conda-forge
-# - bioconda
-# - r
-# - pytorch
-# - defaults
-# Platform: osx-arm64
-# Collecting package metadata (repodata.json): ...working... done
-# Solving environment: ...working... done
-#
-# ## Package Plan ##
-#
-# environment location: /Users/wwang/miniconda3/envs/ONTraC
-#
-# added / updated specs:
-# - python=3.11
-#
-#
-# The following packages will be downloaded:
-#
-# package | build
-# ---------------------------|-----------------
-# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
-# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
-# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
-# ------------------------------------------------------------
-# Total: 4.3 MB
-#
-# The following NEW packages will be INSTALLED:
-#
-# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
-# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
-# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
-# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
-# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
-# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
-# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
-# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
-# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
-# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
-# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
-# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
-# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
-# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
-# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
-# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
-#
-#
-#
-# Downloading and Extracting Packages: ...working... done
-# Preparing transaction: ...working... done
-# Verifying transaction: ...working... done
-# Executing transaction: ...working... done
-# #
-# # To activate this environment, use
-# #
-# # $ conda activate ONTraC
-# #
-# # To deactivate an active environment, use
-# #
-# # $ conda deactivate
-#
-# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
-# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
-# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
-# Running command git checkout -b V2 --track origin/V2
-# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
-# Installing build dependencies: started
-# Switched to a new branch 'V2'
-# branch 'V2' set up to track 'origin/V2'.
-# Installing build dependencies: finished with status 'done'
-# Getting requirements to build wheel: started
-# Getting requirements to build wheel: finished with status 'done'
-# Preparing metadata (pyproject.toml): started
-# Preparing metadata (pyproject.toml): finished with status 'done'
-# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
-# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
-# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
-# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
-# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
-# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
-# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
-# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
-# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
-# Collecting session-info (from ONTraC==2.0a9.dev7)
-# Using cached session_info-1.0.0-py3-none-any.whl
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
-# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
-# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
-# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
-# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
-# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
-# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
-# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
-# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
-# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
-# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
-# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
-# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
-# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
-# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
-# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
-# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
-# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
-# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
-# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
-# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
-# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
-# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
-# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
-# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
-# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
-# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
-# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
-# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
-# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
-# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
-# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
-# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
-# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
-# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
-# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
-# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
-# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
-# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
-# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
-# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
-# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
-# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
-# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
-# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
-# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
-# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
-# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
-# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
-# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
-# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
-# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
-# Using cached idna-3.7-py3-none-any.whl (66 kB)
-# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
-# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
-# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
-# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
-# Building wheels for collected packages: ONTraC
-# Building wheel for ONTraC (pyproject.toml): started
-# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
-# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
-# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
-# Successfully built ONTraC
-# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
-# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
+source ~/.bash_profile
+conda create -y -n ONTraC python=3.11
+conda activate ONTraC
+pip install git+https://github.com/gyuanlab/ONTraC.git@V2
+# Retrieving notices: ...working... done
+# Channels:
+# - conda-forge
+# - bioconda
+# - r
+# - pytorch
+# - defaults
+# Platform: osx-arm64
+# Collecting package metadata (repodata.json): ...working... done
+# Solving environment: ...working... done
+#
+# ## Package Plan ##
+#
+# environment location: /Users/wwang/miniconda3/envs/ONTraC
+#
+# added / updated specs:
+# - python=3.11
+#
+#
+# The following packages will be downloaded:
+#
+# package | build
+# ---------------------------|-----------------
+# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
+# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
+# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
+# ------------------------------------------------------------
+# Total: 4.3 MB
+#
+# The following NEW packages will be INSTALLED:
+#
+# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
+# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
+# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
+# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
+# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
+# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
+# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
+# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
+# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
+# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
+# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
+# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
+# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
+# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
+# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
+# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
+#
+#
+#
+# Downloading and Extracting Packages: ...working... done
+# Preparing transaction: ...working... done
+# Verifying transaction: ...working... done
+# Executing transaction: ...working... done
+# #
+# # To activate this environment, use
+# #
+# # $ conda activate ONTraC
+# #
+# # To deactivate an active environment, use
+# #
+# # $ conda deactivate
+#
+# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
+# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
+# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
+# Running command git checkout -b V2 --track origin/V2
+# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
+# Installing build dependencies: started
+# Switched to a new branch 'V2'
+# branch 'V2' set up to track 'origin/V2'.
+# Installing build dependencies: finished with status 'done'
+# Getting requirements to build wheel: started
+# Getting requirements to build wheel: finished with status 'done'
+# Preparing metadata (pyproject.toml): started
+# Preparing metadata (pyproject.toml): finished with status 'done'
+# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
+# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
+# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
+# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
+# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
+# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
+# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
+# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
+# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
+# Collecting session-info (from ONTraC==2.0a9.dev7)
+# Using cached session_info-1.0.0-py3-none-any.whl
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
+# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
+# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
+# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
+# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
+# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
+# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
+# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
+# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
+# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
+# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
+# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
+# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
+# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
+# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
+# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
+# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
+# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
+# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
+# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
+# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
+# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
+# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
+# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
+# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
+# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
+# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
+# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
+# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
+# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
+# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
+# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
+# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
+# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
+# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
+# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
+# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
+# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
+# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
+# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
+# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
+# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
+# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
+# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
+# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
+# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
+# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
+# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
+# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
+# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
+# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
+# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
+# Using cached idna-3.7-py3-none-any.whl (66 kB)
+# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
+# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
+# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
+# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
+# Building wheels for collected packages: ONTraC
+# Building wheel for ONTraC (pyproject.toml): started
+# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
+# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
+# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
+# Successfully built ONTraC
+# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
+# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
16.1.2.2 Running ONTraC
-
-source ~/.bash_profile
-conda activate ONTraC_v2_py311
-ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
-# ##################################################################################
-#
-# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
-# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
-# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
-# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
-# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
-#
-# version: 2.0a11
-#
-# ##################################################################################
-# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
-# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
-# 11:33:23 --- INFO: -------- I/O options -------
-# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
-# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
-# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
-# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
-# 11:33:23 --- INFO: -------- preprocessing options -------
-# 11:33:23 --- INFO: resolution: 10.0
-# 11:33:23 --- INFO: -------- niche net constr options -------
-# 11:33:23 --- INFO: n_cpu: 4
-# 11:33:23 --- INFO: n_neighbors: 50
-# 11:33:23 --- INFO: n_local: 20
-# 11:33:23 --- INFO: embedding_adjust: False
-# 11:33:23 --- INFO: sigma: 1
-# 11:33:23 --- INFO: -------- train options -------
-# 11:33:23 --- INFO: device: cpu
-# 11:33:23 --- INFO: epochs: 1000
-# 11:33:23 --- INFO: batch_size: 0
-# 11:33:23 --- INFO: patience: 100
-# 11:33:23 --- INFO: min_delta: 0.001
-# 11:33:23 --- INFO: min_epochs: 50
-# 11:33:23 --- INFO: seed: 42
-# 11:33:23 --- INFO: lr: 0.03
-# 11:33:23 --- INFO: hidden_feats: 4
-# 11:33:23 --- INFO: k: 6
-# 11:33:23 --- INFO: modularity_loss_weight: 1.0
-# 11:33:23 --- INFO: purity_loss_weight: 100.0
-# 11:33:23 --- INFO: regularization_loss_weight: 0.1
-# 11:33:23 --- INFO: beta: 0.3
-# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
-# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
-# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
-# 11:33:23 --- INFO: ------------- Niche network construct ---------------
-# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
-# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
-# 11:33:28 --- INFO: Generating samples.yaml file.
-# 11:33:28 --- INFO: ------------ Niche network construct end ------------
-# 11:33:28 --- INFO: ------------------------ GNN ------------------------
-# 11:33:28 --- INFO: Loading dataset.
-# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
-# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
-# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
-# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
-# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
-# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
-# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
-# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
-# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
-# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
-# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
-# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
-# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
-# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
-# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
-# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
-# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
-# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
-# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
-# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
-# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
-# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
-# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
-# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
-# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
-# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
-# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
-# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
-# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
-# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
-# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
-# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
-# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
-# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
-# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
-# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
-# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
-# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
-# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
-# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
-# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
-# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
-# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
-# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
-# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
-# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
-# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
-# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
-# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
-# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
-# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
-# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
-# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
-# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
-# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
-# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
-# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
-# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
-# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
-# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
-# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
-# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
-# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
-# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
-# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
-# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
-# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
-# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
-# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
-# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
-# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
-# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
-# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
-# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
-# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
-# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
-# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
-# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
-# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
-# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
-# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
-# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
-# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
-# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
-# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
-# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
-# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
-# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
-# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
-# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
-# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
-# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
-# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
-# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
-# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
-# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
-# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
-# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
-# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
-# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
-# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
-# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
-# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
-# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
-# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
-# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
-# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
-# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
-# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
-# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
-# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
-# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
-# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
-# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
-# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
-# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
-# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
-# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
-# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
-# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
-# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
-# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
-# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
-# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
-# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
-# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
-# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
-# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
-# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
-# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
-# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
-# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
-# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
-# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
-# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
-# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
-# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
-# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
-# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
-# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
-# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
-# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
-# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
-# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
-# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
-# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
-# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
-# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
-# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
-# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
-# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
-# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
-# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
-# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
-# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
-# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
-# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
-# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
-# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
-# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
-# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
-# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
-# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
-# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
-# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
-# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
-# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
-# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
-# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
-# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
-# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
-# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
-# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
-# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
-# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
-# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
-# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
-# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
-# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
-# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
-# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
-# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
-# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
-# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
-# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
-# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
-# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
-# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
-# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
-# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
-# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
-# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
-# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
-# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
-# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
-# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
-# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
-# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
-# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
-# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
-# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
-# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
-# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
-# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
-# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
-# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
-# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
-# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
-# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
-# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
-# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
-# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
-# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
-# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
-# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
-# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
-# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
-# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
-# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
-# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
-# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
-# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
-# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
-# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
-# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
-# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
-# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
-# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
-# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
-# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
-# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
-# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
-# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
-# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
-# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
-# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
-# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
-# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
-# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
-# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
-# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
-# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
-# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
-# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
-# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
-# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
-# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
-# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
-# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
-# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
-# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
-# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
-# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
-# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
-# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
-# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
-# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
-# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
-# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
-# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
-# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
-# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
-# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
-# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
-# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
-# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
-# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
-# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
-# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
-# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
-# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
-# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
-# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
-# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
-# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
-# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
-# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
-# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
-# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
-# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
-# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
-# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
-# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
-# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
-# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
-# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
-# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
-# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
-# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
-# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
-# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
-# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
-# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
-# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
-# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
-# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
-# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
-# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
-# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
-# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
-# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
-# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
-# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
-# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
-# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
-# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
-# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
-# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
-# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
-# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
-# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
-# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
-# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
-# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
-# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
-# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
-# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
-# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
-# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
-# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
-# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
-# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
-# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
-# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
-# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
-# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
-# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
-# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
-# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
-# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
-# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
-# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
-# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
-# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
-# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
-# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
-# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
-# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
-# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
-# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
-# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
-# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
-# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
-# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
-# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
-# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
-# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
-# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
-# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
-# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
-# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
-# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
-# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
-# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
-# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
-# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
-# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
-# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
-# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
-# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
-# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
-# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
-# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
-# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
-# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
-# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
-# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
-# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
-# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
-# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
-# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
-# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
-# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
-# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
-# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
-# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
-# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
-# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
-# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
-# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
-# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
-# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
-# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
-# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
-# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
-# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
-# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
-# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
-# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
-# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
-# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
-# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
-# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
-# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
-# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
-# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
-# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
-# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
-# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
-# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
-# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
-# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
-# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
-# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
-# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
-# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
-# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
-# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
-# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
-# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
-# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
-# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
-# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
-# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
-# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
-# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
-# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
-# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
-# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
-# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
-# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
-# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
-# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
-# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
-# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
-# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
-# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
-# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
-# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
-# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
-# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
-# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
-# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
-# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
-# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
-# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
-# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
-# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
-# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
-# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
-# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
-# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
-# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
-# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
-# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
-# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
-# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
-# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
-# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
-# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
-# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
-# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
-# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
-# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
-# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
-# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
-# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
-# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
-# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
-# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
-# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
-# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
-# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
-# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
-# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
-# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
-# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
-# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
-# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
-# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
-# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
-# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
-# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
-# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
-# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
-# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
-# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
-# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
-# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
-# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
-# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
-# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
-# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
-# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
-# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
-# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
-# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
-# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
-# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
-# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
-# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
-# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
-# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
-# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
-# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
-# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
-# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
-# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
-# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
-# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
-# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
-# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
-# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
-# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
-# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
-# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
-# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
-# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
-# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
-# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
-# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
-# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
-# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
-# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
-# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
-# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
-# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
-# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
-# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
-# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
-# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
-# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
-# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
-# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
-# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
-# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
-# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
-# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
-# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
-# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
-# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
-# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
-# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
-# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
-# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
-# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
-# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
-# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
-# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
-# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
-# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
-# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
-# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
-# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
-# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
-# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
-# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
-# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
-# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
-# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
-# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
-# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
-# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
-# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
-# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
-# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
-# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
-# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
-# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
-# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
-# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
-# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
-# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
-# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
-# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
-# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
-# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
-# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
-# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
-# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
-# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
-# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
-# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
-# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
-# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
-# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
-# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
-# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
-# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
-# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
-# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
-# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
-# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
-# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
-# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
-# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
-# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
-# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
-# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
-# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
-# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
-# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
-# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
-# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
-# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
-# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
-# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
-# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
-# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
-# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
-# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
-# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
-# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
-# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
-# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
-# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
-# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
-# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
-# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
-# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
-# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
-# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
-# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
-# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
-# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
-# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
-# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
-# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
-# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
-# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
-# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
-# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
-# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
-# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
-# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
-# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
-# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
-# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
-# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
-# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
-# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
-# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
-# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
-# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
-# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
-# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
-# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
-# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
-# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
-# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
-# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
-# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
-# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
-# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
-# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
-# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
-# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
-# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
-# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
-# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
-# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
-# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
-# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
-# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
-# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
-# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
-# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
-# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
-# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
-# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
-# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
-# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
-# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
-# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
-# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
-# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
-# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
-# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
-# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
-# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
-# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
-# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
-# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
-# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
-# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
-# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
-# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
-# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
-# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
-# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
-# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
-# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
-# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
-# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
-# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
-# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
-# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
-# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
-# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
-# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
-# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
-# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
-# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
-# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
-# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
-# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
-# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
-# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
-# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
-# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
-# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
-# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
-# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
-# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
-# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
-# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
-# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
-# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
-# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
-# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
-# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
-# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
-# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
-# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
-# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
-# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
-# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
-# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
-# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
-# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
-# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
-# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
-# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
-# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
-# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
-# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
-# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
-# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
-# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
-# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
-# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
-# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
-# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
-# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
-# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
-# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
-# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
-# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
-# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
-# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
-# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
-# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
-# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
-# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
-# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
-# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
-# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
-# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
-# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
-# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
-# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
-# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
-# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
-# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
-# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
-# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
-# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
-# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
-# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
-# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
-# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
-# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
-# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
-# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
-# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
-# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
-# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
-# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
-# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
-# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
-# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
-# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
-# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
-# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
-# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
-# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
-# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
-# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
-# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
-# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
-# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
-# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
-# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
-# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
-# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
-# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
-# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
-# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
-# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
-# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
-# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
-# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
-# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
-# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
-# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
-# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
-# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
-# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
-# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
-# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
-# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
-# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
-# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
-# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
-# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
-# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
-# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
-# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
-# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
-# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
-# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
-# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
-# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
-# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
-# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
-# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
-# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
-# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
-# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
-# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
-# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
-# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
-# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
-# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
-# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
-# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
-# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
-# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
-# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
-# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
-# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
-# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
-# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
-# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
-# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
-# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
-# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
-# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
-# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
-# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
-# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
-# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
-# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
-# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
-# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
-# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
-# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
-# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
-# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
-# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
-# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
-# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
-# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
-# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
-# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
-# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
-# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
-# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
-# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
-# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
-# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
-# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
-# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
-# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
-# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
-# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
-# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
-# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
-# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
-# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
-# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
-# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
-# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
-# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
-# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
-# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
-# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
-# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
-# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
-# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
-# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
-# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
-# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
-# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
-# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
-# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
-# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
-# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
-# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
-# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
-# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
-# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
-# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
-# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
-# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
-# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
-# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
-# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
-# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
-# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
-# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
-# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
-# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
-# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
-# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
-# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
-# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
-# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
-# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
-# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
-# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
-# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
-# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
-# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
-# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
-# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
-# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
-# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
-# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
-# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
-# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
-# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
-# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
-# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
-# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
-# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
-# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
-# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
-# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
-# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
-# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
-# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
-# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
-# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
-# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
-# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
-# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
-# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
-# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
-# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
-# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
-# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
-# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
-# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
-# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
-# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
-# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
-# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
-# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
-# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
-# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
-# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
-# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
-# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
-# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
-# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
-# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
-# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
-# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
-# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
-# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
-# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
-# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
-# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
-# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
-# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
-# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
-# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
-# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
-# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
-# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
-# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
-# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
-# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
-# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
-# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
-# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
-# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
-# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
-# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
-# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
-# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
-# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
-# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
-# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
-# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
-# 11:38:37 --- INFO: Training process end.
-# 11:38:37 --- INFO: Evaluating process start.
-# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
-# 11:38:37 --- INFO: Evaluating process end.
-# 11:38:37 --- INFO: Predicting process start.
-# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
-# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
-# 11:38:38 --- INFO: Predicting process end.
-# 11:38:38 --- INFO: --------------------- GNN end ----------------------
-# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
-# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
-# 11:38:38 --- INFO: Loading dataset.
-# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:38:39 --- INFO: Calculating NTScore for each niche.
-# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
-# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
-# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
-# 11:38:39 --- INFO: Output NTScore tables.
-# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
+
+source ~/.bash_profile
+conda activate ONTraC_v2_py311
+ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
+# ##################################################################################
+#
+# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
+# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
+# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
+# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
+# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
+#
+# version: 2.0a11
+#
+# ##################################################################################
+# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
+# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
+# 11:33:23 --- INFO: -------- I/O options -------
+# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
+# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
+# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
+# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
+# 11:33:23 --- INFO: -------- preprocessing options -------
+# 11:33:23 --- INFO: resolution: 10.0
+# 11:33:23 --- INFO: -------- niche net constr options -------
+# 11:33:23 --- INFO: n_cpu: 4
+# 11:33:23 --- INFO: n_neighbors: 50
+# 11:33:23 --- INFO: n_local: 20
+# 11:33:23 --- INFO: embedding_adjust: False
+# 11:33:23 --- INFO: sigma: 1
+# 11:33:23 --- INFO: -------- train options -------
+# 11:33:23 --- INFO: device: cpu
+# 11:33:23 --- INFO: epochs: 1000
+# 11:33:23 --- INFO: batch_size: 0
+# 11:33:23 --- INFO: patience: 100
+# 11:33:23 --- INFO: min_delta: 0.001
+# 11:33:23 --- INFO: min_epochs: 50
+# 11:33:23 --- INFO: seed: 42
+# 11:33:23 --- INFO: lr: 0.03
+# 11:33:23 --- INFO: hidden_feats: 4
+# 11:33:23 --- INFO: k: 6
+# 11:33:23 --- INFO: modularity_loss_weight: 1.0
+# 11:33:23 --- INFO: purity_loss_weight: 100.0
+# 11:33:23 --- INFO: regularization_loss_weight: 0.1
+# 11:33:23 --- INFO: beta: 0.3
+# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
+# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
+# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
+# 11:33:23 --- INFO: ------------- Niche network construct ---------------
+# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
+# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
+# 11:33:28 --- INFO: Generating samples.yaml file.
+# 11:33:28 --- INFO: ------------ Niche network construct end ------------
+# 11:33:28 --- INFO: ------------------------ GNN ------------------------
+# 11:33:28 --- INFO: Loading dataset.
+# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
+# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
+# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
+# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
+# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
+# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
+# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
+# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
+# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
+# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
+# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
+# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
+# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
+# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
+# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
+# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
+# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
+# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
+# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
+# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
+# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
+# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
+# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
+# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
+# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
+# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
+# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
+# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
+# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
+# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
+# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
+# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
+# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
+# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
+# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
+# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
+# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
+# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
+# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
+# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
+# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
+# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
+# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
+# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
+# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
+# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
+# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
+# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
+# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
+# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
+# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
+# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
+# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
+# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
+# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
+# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
+# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
+# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
+# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
+# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
+# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
+# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
+# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
+# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
+# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
+# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
+# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
+# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
+# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
+# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
+# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
+# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
+# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
+# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
+# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
+# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
+# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
+# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
+# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
+# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
+# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
+# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
+# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
+# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
+# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
+# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
+# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
+# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
+# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
+# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
+# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
+# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
+# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
+# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
+# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
+# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
+# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
+# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
+# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
+# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
+# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
+# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
+# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
+# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
+# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
+# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
+# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
+# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
+# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
+# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
+# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
+# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
+# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
+# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
+# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
+# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
+# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
+# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
+# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
+# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
+# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
+# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
+# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
+# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
+# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
+# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
+# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
+# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
+# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
+# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
+# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
+# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
+# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
+# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
+# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
+# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
+# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
+# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
+# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
+# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
+# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
+# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
+# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
+# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
+# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
+# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
+# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
+# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
+# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
+# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
+# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
+# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
+# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
+# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
+# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
+# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
+# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
+# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
+# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
+# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
+# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
+# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
+# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
+# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
+# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
+# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
+# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
+# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
+# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
+# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
+# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
+# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
+# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
+# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
+# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
+# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
+# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
+# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
+# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
+# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
+# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
+# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
+# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
+# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
+# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
+# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
+# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
+# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
+# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
+# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
+# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
+# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
+# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
+# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
+# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
+# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
+# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
+# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
+# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
+# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
+# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
+# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
+# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
+# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
+# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
+# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
+# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
+# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
+# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
+# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
+# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
+# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
+# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
+# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
+# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
+# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
+# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
+# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
+# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
+# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
+# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
+# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
+# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
+# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
+# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
+# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
+# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
+# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
+# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
+# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
+# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
+# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
+# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
+# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
+# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
+# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
+# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
+# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
+# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
+# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
+# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
+# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
+# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
+# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
+# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
+# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
+# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
+# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
+# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
+# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
+# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
+# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
+# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
+# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
+# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
+# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
+# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
+# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
+# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
+# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
+# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
+# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
+# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
+# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
+# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
+# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
+# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
+# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
+# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
+# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
+# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
+# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
+# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
+# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
+# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
+# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
+# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
+# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
+# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
+# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
+# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
+# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
+# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
+# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
+# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
+# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
+# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
+# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
+# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
+# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
+# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
+# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
+# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
+# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
+# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
+# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
+# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
+# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
+# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
+# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
+# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
+# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
+# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
+# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
+# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
+# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
+# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
+# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
+# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
+# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
+# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
+# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
+# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
+# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
+# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
+# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
+# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
+# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
+# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
+# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
+# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
+# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
+# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
+# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
+# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
+# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
+# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
+# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
+# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
+# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
+# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
+# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
+# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
+# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
+# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
+# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
+# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
+# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
+# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
+# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
+# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
+# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
+# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
+# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
+# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
+# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
+# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
+# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
+# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
+# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
+# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
+# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
+# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
+# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
+# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
+# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
+# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
+# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
+# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
+# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
+# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
+# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
+# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
+# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
+# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
+# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
+# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
+# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
+# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
+# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
+# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
+# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
+# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
+# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
+# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
+# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
+# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
+# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
+# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
+# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
+# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
+# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
+# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
+# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
+# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
+# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
+# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
+# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
+# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
+# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
+# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
+# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
+# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
+# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
+# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
+# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
+# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
+# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
+# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
+# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
+# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
+# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
+# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
+# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
+# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
+# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
+# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
+# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
+# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
+# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
+# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
+# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
+# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
+# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
+# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
+# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
+# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
+# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
+# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
+# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
+# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
+# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
+# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
+# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
+# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
+# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
+# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
+# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
+# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
+# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
+# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
+# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
+# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
+# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
+# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
+# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
+# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
+# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
+# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
+# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
+# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
+# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
+# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
+# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
+# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
+# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
+# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
+# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
+# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
+# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
+# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
+# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
+# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
+# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
+# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
+# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
+# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
+# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
+# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
+# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
+# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
+# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
+# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
+# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
+# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
+# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
+# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
+# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
+# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
+# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
+# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
+# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
+# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
+# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
+# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
+# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
+# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
+# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
+# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
+# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
+# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
+# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
+# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
+# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
+# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
+# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
+# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
+# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
+# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
+# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
+# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
+# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
+# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
+# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
+# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
+# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
+# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
+# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
+# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
+# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
+# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
+# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
+# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
+# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
+# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
+# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
+# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
+# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
+# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
+# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
+# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
+# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
+# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
+# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
+# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
+# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
+# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
+# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
+# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
+# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
+# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
+# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
+# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
+# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
+# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
+# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
+# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
+# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
+# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
+# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
+# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
+# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
+# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
+# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
+# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
+# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
+# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
+# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
+# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
+# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
+# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
+# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
+# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
+# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
+# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
+# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
+# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
+# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
+# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
+# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
+# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
+# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
+# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
+# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
+# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
+# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
+# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
+# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
+# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
+# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
+# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
+# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
+# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
+# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
+# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
+# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
+# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
+# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
+# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
+# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
+# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
+# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
+# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
+# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
+# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
+# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
+# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
+# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
+# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
+# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
+# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
+# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
+# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
+# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
+# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
+# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
+# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
+# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
+# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
+# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
+# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
+# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
+# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
+# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
+# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
+# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
+# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
+# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
+# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
+# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
+# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
+# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
+# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
+# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
+# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
+# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
+# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
+# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
+# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
+# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
+# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
+# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
+# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
+# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
+# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
+# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
+# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
+# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
+# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
+# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
+# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
+# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
+# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
+# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
+# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
+# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
+# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
+# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
+# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
+# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
+# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
+# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
+# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
+# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
+# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
+# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
+# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
+# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
+# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
+# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
+# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
+# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
+# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
+# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
+# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
+# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
+# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
+# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
+# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
+# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
+# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
+# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
+# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
+# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
+# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
+# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
+# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
+# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
+# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
+# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
+# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
+# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
+# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
+# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
+# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
+# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
+# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
+# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
+# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
+# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
+# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
+# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
+# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
+# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
+# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
+# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
+# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
+# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
+# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
+# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
+# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
+# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
+# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
+# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
+# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
+# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
+# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
+# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
+# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
+# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
+# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
+# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
+# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
+# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
+# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
+# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
+# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
+# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
+# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
+# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
+# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
+# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
+# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
+# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
+# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
+# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
+# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
+# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
+# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
+# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
+# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
+# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
+# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
+# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
+# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
+# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
+# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
+# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
+# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
+# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
+# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
+# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
+# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
+# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
+# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
+# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
+# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
+# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
+# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
+# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
+# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
+# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
+# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
+# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
+# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
+# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
+# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
+# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
+# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
+# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
+# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
+# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
+# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
+# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
+# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
+# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
+# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
+# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
+# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
+# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
+# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
+# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
+# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
+# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
+# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
+# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
+# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
+# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
+# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
+# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
+# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
+# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
+# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
+# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
+# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
+# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
+# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
+# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
+# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
+# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
+# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
+# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
+# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
+# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
+# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
+# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
+# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
+# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
+# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
+# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
+# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
+# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
+# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
+# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
+# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
+# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
+# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
+# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
+# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
+# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
+# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
+# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
+# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
+# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
+# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
+# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
+# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
+# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
+# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
+# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
+# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
+# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
+# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
+# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
+# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
+# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
+# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
+# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
+# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
+# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
+# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
+# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
+# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
+# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
+# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
+# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
+# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
+# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
+# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
+# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
+# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
+# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
+# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
+# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
+# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
+# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
+# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
+# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
+# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
+# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
+# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
+# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
+# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
+# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
+# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
+# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
+# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
+# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
+# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
+# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
+# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
+# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
+# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
+# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
+# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
+# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
+# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
+# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
+# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
+# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
+# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
+# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
+# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
+# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
+# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
+# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
+# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
+# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
+# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
+# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
+# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
+# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
+# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
+# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
+# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
+# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
+# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
+# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
+# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
+# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
+# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
+# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
+# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
+# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
+# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
+# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
+# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
+# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
+# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
+# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
+# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
+# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
+# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
+# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
+# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
+# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
+# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
+# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
+# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
+# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
+# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
+# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
+# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
+# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
+# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
+# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
+# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
+# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
+# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
+# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
+# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
+# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
+# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
+# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
+# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
+# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
+# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
+# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
+# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
+# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
+# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
+# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
+# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
+# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
+# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
+# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
+# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
+# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
+# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
+# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
+# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
+# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
+# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
+# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
+# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
+# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
+# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
+# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
+# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
+# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
+# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
+# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
+# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
+# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
+# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
+# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
+# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
+# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
+# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
+# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
+# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
+# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
+# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
+# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
+# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
+# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
+# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
+# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
+# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
+# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
+# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
+# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
+# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
+# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
+# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
+# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
+# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
+# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
+# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
+# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
+# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
+# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
+# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
+# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
+# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
+# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
+# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
+# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
+# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
+# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
+# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
+# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
+# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
+# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
+# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
+# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
+# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
+# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
+# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
+# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
+# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
+# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
+# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
+# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
+# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
+# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
+# 11:38:37 --- INFO: Training process end.
+# 11:38:37 --- INFO: Evaluating process start.
+# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
+# 11:38:37 --- INFO: Evaluating process end.
+# 11:38:37 --- INFO: Predicting process start.
+# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
+# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
+# 11:38:38 --- INFO: Predicting process end.
+# 11:38:38 --- INFO: --------------------- GNN end ----------------------
+# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
+# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
+# 11:38:38 --- INFO: Loading dataset.
+# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:38:39 --- INFO: Calculating NTScore for each niche.
+# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
+# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
+# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
+# 11:38:39 --- INFO: Output NTScore tables.
+# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
16.1.3 visualization
16.1.3.1 load ONTraC results
-
+
The NTScore and binarized niche cluster info were stored in cell metadata
-
-# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
-# <char> <char> <char> <char> <char> <char> <char> <int> <num>
-# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
-# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
-# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
-# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
-# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
-# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
+
+# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
+# <char> <char> <char> <char> <char> <char> <char> <int> <num>
+# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
+# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
+# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
+# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
+# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
+# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
The probability matrix of each cell assigned to each niche cluster and
connectivity between niche cluster were stored here.
-
-
+
+
16.1.3.2 niche cluster prob
-spatFeatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'niche cluster',
- expression_values = "prob",
- group_by = 'list_ID',
- feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
- point_border_col = 'gray'
- )
+spatFeatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'niche cluster',
+ expression_values = "prob",
+ group_by = 'list_ID',
+ feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
+ point_border_col = 'gray'
+ )
16.1.3.3 binarized niche cluster
-spatPlot2D(giotto_obj,
- spat_unit = "cell",
- group_by = 'list_ID',
- cell_color = "NicheCluster",
- color_as_factor = T,
- point_size = 1,
- point_border_stroke = NA,
- title="Niche cluster")
+spatPlot2D(giotto_obj,
+ spat_unit = "cell",
+ group_by = 'list_ID',
+ cell_color = "NicheCluster",
+ color_as_factor = T,
+ point_size = 1,
+ point_border_stroke = NA,
+ title="Niche cluster")
16.1.3.5 NT score
-spatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'rna',
- group_by = 'list_ID',
- cell_color = "NTScore",
- color_as_factor = FALSE,
- cell_color_gradient = "turbo",
- point_size = 1,
- point_border_stroke = NA
- )
+spatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ group_by = 'list_ID',
+ cell_color = "NTScore",
+ color_as_factor = FALSE,
+ cell_color_gradient = "turbo",
+ point_size = 1,
+ point_border_stroke = NA
+ )
-
+
diff --git a/interoperability-with-other-frameworks.html b/interoperability-with-other-frameworks.html
index 7050533..347d3af 100644
--- a/interoperability-with-other-frameworks.html
+++ b/interoperability-with-other-frameworks.html
@@ -521,10 +521,10 @@ 15 Interoperability with other fr
15.1 Load Giotto object
To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process:
-
+
Load a Giotto mini Visium object, which will be used for demonstrating interoperability.
-
+
15.2 Seurat
@@ -533,99 +533,99 @@ 15.2 Seurat
15.2.1 Conversion of Giotto Object to Seurat Object
To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function
-
+
Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data.
-gToS <- giottoToSeuratV5(gobject = gobject,
- spat_unit = "cell")
-
-plot1 <- VlnPlot(gToS,
- features = "nCount_rna",
- pt.size = 0.1) +
- NoLegend()
-
-plot2 <- SpatialFeaturePlot(gToS,
- features = "nCount_rna",
- pt.size.factor = 2) +
- theme(legend.position = "right")
-
-wrap_plots(plot1, plot2)
+gToS <- giottoToSeuratV5(gobject = gobject,
+ spat_unit = "cell")
+
+plot1 <- VlnPlot(gToS,
+ features = "nCount_rna",
+ pt.size = 0.1) +
+ NoLegend()
+
+plot2 <- SpatialFeaturePlot(gToS,
+ features = "nCount_rna",
+ pt.size.factor = 2) +
+ theme(legend.position = "right")
+
+wrap_plots(plot1, plot2)
15.2.1.1 Apply SCTransform
We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function.
-
+
15.2.1.2 Dimension Reduction
We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object:
-
+
15.2.2 Conversion of Seurat object Back to Giotto Object
To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks.
-giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
- spatial_assay = "rna",
- dim_reduction = c("pca", "umap"),
- sp_network = "Delaunay_network",
- nn_network = c("sNN.pca", "custom_NN" ))
+giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
+ spatial_assay = "rna",
+ dim_reduction = c("pca", "umap"),
+ sp_network = "Delaunay_network",
+ nn_network = c("sNN.pca", "custom_NN" ))
15.2.2.1 Clustering and Plotting UMAP
Here we perform K-means clustering on the UMAP results obtained from the Seurat object:
-## k-means clustering
-giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
- dim_reduction_to_use = "pca")
-
-#Plot UMAP post-clustering to visualize kmeans
-graph2 <- Giotto::plotUMAP(
- gobject = giottoFromSeurat,
- cell_color = "kmeans",
- show_NN_network = TRUE,
- point_size = 2.5
-)
+## k-means clustering
+giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
+ dim_reduction_to_use = "pca")
+
+#Plot UMAP post-clustering to visualize kmeans
+graph2 <- Giotto::plotUMAP(
+ gobject = giottoFromSeurat,
+ cell_color = "kmeans",
+ show_NN_network = TRUE,
+ point_size = 2.5
+)
15.2.2.2 Spatial CoExpression
We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function:
-ranktest <- binSpect(giottoFromSeurat,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "Delaunay_network")
-
-ext_spatial_genes <- ranktest[1:300,]$feats
-
-spat_cor_netw_DT <- detectSpatialCorFeats(
- giottoFromSeurat,
- method = "network",
- spatial_network_name = "Delaunay_network",
- subset_feats = ext_spatial_genes)
-
-top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
- feats = "Dsp",
- show_top_feats = 10)
-
-spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
- name = "spat_netw_clus",
- k = 7)
-heatmSpatialCorFeats(
- giottoFromSeurat,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL),
- save_plot = TRUE,
- show_plot = TRUE,
- return_plot = FALSE,
- save_param = list(base_height = 6, base_width = 8, units = 'cm'))
+ranktest <- binSpect(giottoFromSeurat,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "Delaunay_network")
+
+ext_spatial_genes <- ranktest[1:300,]$feats
+
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ giottoFromSeurat,
+ method = "network",
+ spatial_network_name = "Delaunay_network",
+ subset_feats = ext_spatial_genes)
+
+top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
+ feats = "Dsp",
+ show_top_feats = 10)
+
+spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
+ name = "spat_netw_clus",
+ k = 7)
+heatmSpatialCorFeats(
+ giottoFromSeurat,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ return_plot = FALSE,
+ save_param = list(base_height = 6, base_width = 8, units = 'cm'))
@@ -635,60 +635,60 @@ 15.3 SpatialExperiment
To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries.
-library(SpatialExperiment)
-library(ggspavis)
-library(pheatmap)
-library(scater)
-library(scran)
-library(nnSVG)
+library(SpatialExperiment)
+library(ggspavis)
+library(pheatmap)
+library(scater)
+library(scran)
+library(nnSVG)
15.3.1 Convert Giotto Object to SpatialExperiment Object
To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function.
-
+
The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use:
-
+
15.3.1.1 Identify top spatially variable genes with nnSVG
We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate:
-# One of the assays should be "logcounts"
-# We rename the normalized assay to "logcounts"
-assayNames(spe)[[2]] <- "logcounts"
-
-# Create model matrix for leiden clustering labels
-X <- model.matrix(~ colData(spe)$leiden_clus)
-dim(X)
+# One of the assays should be "logcounts"
+# We rename the normalized assay to "logcounts"
+assayNames(spe)[[2]] <- "logcounts"
+
+# Create model matrix for leiden clustering labels
+X <- model.matrix(~ colData(spe)$leiden_clus)
+dim(X)
- Run nnSVG
This step will take several minutes to run
-
+
15.3.2 Conversion of SpatialExperiment object back to Giotto
We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object.
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-print(giottoFromSPE)
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+print(giottoFromSPE)
15.3.2.1 Plotting top genes from nnSVG in Giotto
Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object.
-ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
- output = "data.table")
-
-ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
-spatFeatPlot2D(giottoFromSPE,
- expression_values = "scaled_rna_cell",
- feats = ext_spatial_genes[1:4],
- point_size = 2)
+ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
+ output = "data.table")
+
+ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
+spatFeatPlot2D(giottoFromSPE,
+ expression_values = "scaled_rna_cell",
+ feats = ext_spatial_genes[1:4],
+ point_size = 2)
@@ -701,97 +701,97 @@ 15.4 AnnData
15.4.1 Load Required Libraries
To start, we need to load the necessary libraries, including reticulate for interfacing with Python.
-
+
15.4.2 Specify Path for Results
First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions.
-# Specify path to which results may be saved
-results_directory <- "results/03_session4/giotto_anndata_conversion/"
-
-instrs <- showGiottoInstructions(gobject)
-
-mini_gobject <- replaceGiottoInstructions(gobject = gobject,
- instructions = instrs)
+# Specify path to which results may be saved
+results_directory <- "results/03_session4/giotto_anndata_conversion/"
+
+instrs <- showGiottoInstructions(gobject)
+
+mini_gobject <- replaceGiottoInstructions(gobject = gobject,
+ instructions = instrs)
15.4.2.1 Create Default kNN Network
We will create a k-nearest neighbor (kNN) network using mostly default parameters.
-
+
15.4.3 Giotto To AnnData
To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData()
-
+
Next, we import scanpy and perform a series of preprocessing steps on the AnnData object.
-scanpy <- import("scanpy")
-
-adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
-
-# Normalize total counts per cell
-scanpy$pp$normalize_total(adata, target_sum=1e4)
-
-# Log-transform the data
-scanpy$pp$log1p(adata)
-
-# Perform PCA
-scanpy$pp$pca(adata, n_comps=40L)
-
-# Compute the neighborhood graph
-scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
-
-# Run UMAP
-scanpy$tl$umap(adata)
-
-# Save the processed AnnData object
-adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
-
-processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
+scanpy <- import("scanpy")
+
+adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
+
+# Normalize total counts per cell
+scanpy$pp$normalize_total(adata, target_sum=1e4)
+
+# Log-transform the data
+scanpy$pp$log1p(adata)
+
+# Perform PCA
+scanpy$pp$pca(adata, n_comps=40L)
+
+# Compute the neighborhood graph
+scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
+
+# Run UMAP
+scanpy$tl$umap(adata)
+
+# Save the processed AnnData object
+adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
+
+processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
15.4.4 Convert AnnData to Giotto
Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto.
-
+
15.4.4.1 UMAP Visualization
Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object.
-Giotto::plotUMAP(
- gobject = giottoFromAnndata,
- dim_reduction_name = "umap.ad",
- cell_color = "leiden_clus",
- point_size = 3
-)
+Giotto::plotUMAP(
+ gobject = giottoFromAnndata,
+ dim_reduction_name = "umap.ad",
+ cell_color = "leiden_clus",
+ point_size = 3
+)
15.5 Create mini Vizgen object
-mini_gobject <- loadGiottoMini(dataset = "vizgen",
- python_path = NULL)
-
-mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
- instructions = instrs)
-mini_gobject <- createNearestNetwork(gobject = mini_gobject,
- spat_unit = "aggregate",
- feat_type = "rna",
- type = "kNN",
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap",
- k = 6,
- name = "new_network")
+mini_gobject <- loadGiottoMini(dataset = "vizgen",
+ python_path = NULL)
+
+mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
+ instructions = instrs)
+mini_gobject <- createNearestNetwork(gobject = mini_gobject,
+ spat_unit = "aggregate",
+ feat_type = "rna",
+ type = "kNN",
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap",
+ k = 6,
+ name = "new_network")
Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details.
-
+
diff --git a/introduction-to-the-giotto-package.html b/introduction-to-the-giotto-package.html
index e22d6d1..0ab2014 100644
--- a/introduction-to-the-giotto-package.html
+++ b/introduction-to-the-giotto-package.html
@@ -546,19 +546,58 @@ 4.3 Installation + python environ
4.3.1 Giotto installation
Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository.
Much of this already covered in Section 2.2, but the highlights are:
-
-- Ensure your system
-
-To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak.
+
+4.3.1.1 System prerequisites
+
+- for windows, Rtools needs to be installed
+- a major dependency terra needs
GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux
+
+
+
+4.3.1.2 Installation of released version
+To install the currently released version of Giotto in a single step:
+This should automatically install all the Giotto dependencies and other Giotto module packages.
+
+
+4.3.1.3 Installation of dev branch Giotto packages
+You can install dev branch versions by using devtools::install_github()
+Core module dev branchs:
+
+- “drieslab/Giotto@suite_dev”
+- “drieslab/GiottoVisuals@dev”
+- “drieslab/GiottoClass@dev”
+- “drieslab/GiottoUtils@dev”
+
+
+pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages.
+
+
+4.3.1.4 Common install issues
+If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work.
+
+If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it.
+
+## # Downloading packages -------------------------------------------------------
+## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s]
+## Successfully downloaded 1 package in 1.2 seconds.
+##
+## The following package(s) will be installed:
+## - irlba [2.3.5.1]
+## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu".
+##
+## # Installing packages --------------------------------------------------------
+## - Installing irlba ... OK [built from source and cached in 5.7s]
+## Successfully installed 1 package in 5.7 seconds.
+
4.3.2 Python environment
4.3.2.1 Default installation
In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment.
-
+
Two things are needed for python to work:
- A conda (e.g. miniconda or anaconda) installation which is the package and environment management system.
@@ -580,17 +619,17 @@ 4.3.2.1 Default installation
4.3.2.2 Custom installs
Custom python environments can be made by first setting up a new environment and establishing the name and python version to use.
-
+
Following that, one or more python packages to install can be added to the environment.
-reticulate::py_install(
- pip = TRUE,
- envname = '[name of env]',
- packages = c(
- "package1",
- "package2",
- "..."
- )
-)
+reticulate::py_install(
+ pip = TRUE,
+ envname = '[name of env]',
+ packages = c(
+ "package1",
+ "package2",
+ "..."
+ )
+)
Once an environment has been set up, Giotto can hook into it.
@@ -612,17 +651,56 @@ 4.3.2.3 Using a specific environm
Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto.
You would run file.edit("~/.Rprofile") and then add options("giotto.py_path" = "[path to env or envname]") as a line so that it is automatically set at the start of each session.
If a specific environment should only be used a couple times then method 1 is easiest:
-
+
To check which conda environments exist on your machine:
-
+
Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running:
-
+
4.4 Giotto instructions
-Giotto
+Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are:
+
+- python_path - when set, will activate a python environment
+- save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render.
+- save_plot - whether to save plots to the
save_dir
+- return_plot - whether to return the plot objects. When FALSE, only NULL is returned
+- show_plot - whether to show the plot in the viewer
+
+These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function.
+library(Giotto)
+save_dir <- "results/01_session2/"
+
+# this call will also intialize the python env
+instrs <- createGiottoInstructions(
+ save_dir = save_dir, # working directory is the default
+ show_plot = FALSE,
+ save_plot = TRUE,
+ return_plot = FALSE,
+ python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default
+)
+force(instrs)
+Giotto object creation functions all have an instructions param for passing in instructions objects.
+giotto objects will also respond to the instructions() generic.
+test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object
+
+# example plot
+g <- GiottoData::loadGiottoMini("visium")
+instructions(g) <- instrs
+instructions(g, "show_plot") # instructions say not to plot to viewer
+spatPlot2D(g, show_image = TRUE, image_name = "image")
+# instead it will directly write to the results folder
+As an example, you can also set individual instructions
+
+
+
+
+Figure 4.2: example image output
+
+
diff --git a/multi-omics-integration.html b/multi-omics-integration.html
index 47cad23..e76e25a 100644
--- a/multi-omics-integration.html
+++ b/multi-omics-integration.html
@@ -520,14 +520,14 @@ 14 Multi-omics integration
14.1 The CytAssist technology
The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image.
-
+
Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics.
The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument.
-
+
Figure 14.2: CytAssist workflow. Source: 10X genomics.
@@ -542,19 +542,19 @@
14.2 Introduction to the spatial
14.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/03_session3")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
+dir.create("data/03_session3")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”.
-
+
14.4 Create the Giotto object
@@ -564,124 +564,124 @@ 14.4 Create the Giotto objectx,y(,z) coordinates for cells or spots (or the path to)
createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/03_session3/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-# Provide the path to the data folder
-data_path <- "data/03_session3/"
-
-# Create object directly from the data folder
-visium_tonsil <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/03_session3/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+# Provide the path to the data folder
+data_path <- "data/03_session3/"
+
+# Create object directly from the data folder
+visium_tonsil <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
- Print the information of the object, note that both rna and protein are listed in the expression slot.
-
+
14.5 Subset on spots that were covered by tissue
-spatPlot2D(
- gobject = visium_tonsil,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE,
- image_name = "image"
-)
-
+spatPlot2D(
+ gobject = visium_tonsil,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE,
+ image_name = "image"
+)
+
Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object.
-
+
14.6 RNA processing
- Run the Filtering, normalization, and statistics steps using only the RNA feature.
-visium_tonsil <- filterGiotto(
- gobject = visium_tonsil,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil)
+visium_tonsil <- filterGiotto(
+ gobject = visium_tonsil,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1000,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil)
- Dimension reduction
Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs.
-visium_tonsil <- calculateHVF(gobject = visium_tonsil)
-
-visium_tonsil <- runPCA(gobject = visium_tonsil)
+visium_tonsil <- calculateHVF(gobject = visium_tonsil)
+
+visium_tonsil <- runPCA(gobject = visium_tonsil)
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the RNA-based Leiden clusters.
-
+
- Visualization
Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot.
-plotUMAP(gobject = visium_tonsil,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters.
Plot the spatial distribution of the RNA-based Leiden cluster per spot.
-spatPlot2D(gobject = visium_tonsil,
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.5: Spatial distribution of RNA-based Leiden clusters.
@@ -693,80 +693,80 @@
14.7 Protein processingvisium_tonsil <- filterGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein")
+visium_tonsil <- filterGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein")
- Dimension reduction
Calculate the principal components using all the proteins available in the dataset.
-
+
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the Protein-based Leiden clusters.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- resolution = 1,
- n_iterations = 1000)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ resolution = 1,
+ n_iterations = 1000)
- Visualization
Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters.
Plot the spatial distribution of the Protein-based Leiden clusters.
-spatPlot2D(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.7: Spatial distribution of Protein-based Leiden clusters.
@@ -779,68 +779,68 @@
14.8 Multi-omics integrationCalculate kNN
Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type.
-## RNA modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
-
-## Protein modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
+## RNA modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
+
+## Protein modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
- Run WNN
Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object.
-visium_tonsil <- runWNN(visium_tonsil,
- spat_unit = "cell",
- modality_1 = "rna",
- modality_2 = "protein",
- pca_name_modality_1 = "pca",
- pca_name_modality_2 = "protein.pca",
- k = 20,
- integrated_feat_type = NULL,
- matrix_result_name = NULL,
- w_name_modality_1 = NULL,
- w_name_modality_2 = NULL,
- verbose = TRUE)
+visium_tonsil <- runWNN(visium_tonsil,
+ spat_unit = "cell",
+ modality_1 = "rna",
+ modality_2 = "protein",
+ pca_name_modality_1 = "pca",
+ pca_name_modality_2 = "protein.pca",
+ k = 20,
+ integrated_feat_type = NULL,
+ matrix_result_name = NULL,
+ w_name_modality_1 = NULL,
+ w_name_modality_2 = NULL,
+ verbose = TRUE)
- Run Integrated umap
Calculate the UMAP using the weights of each feature per spot.
-visium_tonsil <- runIntegratedUMAP(visium_tonsil,
- modality1 = "rna",
- modality2 = "protein",
- spread = 7,
- min_dist = 1,
- force = FALSE)
+visium_tonsil <- runIntegratedUMAP(visium_tonsil,
+ modality1 = "rna",
+ modality2 = "protein",
+ spread = 7,
+ min_dist = 1,
+ force = FALSE)
- Calculate integrated clusters
Calculate the multiomics-based Leiden clusters using the weights of each feature per spot.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- nn_network_to_use = "kNN",
- network_name = "integrated_kNN",
- name = "integrated_leiden_clus",
- resolution = 1)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ nn_network_to_use = "kNN",
+ network_name = "integrated_kNN",
+ name = "integrated_leiden_clus",
+ resolution = 1)
- Visualize the integrated UMAP
Plot the integrated UMAP and color the spots using the integrated Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- dim_reduction_name = "integrated.umap",
- point_size = 2,
- title = "Integrated UMAP using Integrated Leiden clusters")
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ dim_reduction_name = "integrated.umap",
+ point_size = 2,
+ title = "Integrated UMAP using Integrated Leiden clusters")
+
Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters.
@@ -850,14 +850,14 @@
14.8 Multi-omics integrationVisualize spatial plot with integrated clusters
Plot the spatial distribution of the integrated Leiden clusters.
-spatPlot2D(visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- point_size = 3,
- show_image = TRUE,
- title = "Integrated Leiden clustering")
-
+spatPlot2D(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ point_size = 3,
+ show_image = TRUE,
+ title = "Integrated Leiden clustering")
+
Figure 14.9: Spatial distribution of the integrated Leiden clusters.
@@ -866,85 +866,85 @@
14.8 Multi-omics integration
14.9 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
- [7] matrixStats_1.3.0 compiler_4.4.1
- [9] png_0.1-8 systemfonts_1.1.0
- [11] vctrs_0.6.5 reshape2_1.4.4
- [13] stringr_1.5.1 pkgconfig_2.0.3
- [15] SpatialExperiment_1.14.0 crayon_1.5.3
- [17] fastmap_1.2.0 backports_1.5.0
- [19] magick_2.8.4 XVector_0.44.0
- [21] labeling_0.4.3 utf8_1.2.4
- [23] rmarkdown_2.27 UCSC.utils_1.0.0
- [25] ragg_1.3.2 purrr_1.0.2
- [27] xfun_0.46 beachmat_2.20.0
- [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
- [31] jsonlite_1.8.8 DelayedArray_0.30.1
- [33] BiocParallel_1.38.0 terra_1.7-78
- [35] irlba_2.3.5.1 parallel_4.4.1
- [37] R6_2.5.1 stringi_1.8.4
- [39] RColorBrewer_1.1-3 reticulate_1.38.0
- [41] parallelly_1.37.1 GenomicRanges_1.56.1
- [43] scattermore_1.2 Rcpp_1.0.13
- [45] bookdown_0.40 SummarizedExperiment_1.34.0
- [47] knitr_1.48 future.apply_1.11.2
- [49] R.utils_2.12.3 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
+ [7] matrixStats_1.3.0 compiler_4.4.1
+ [9] png_0.1-8 systemfonts_1.1.0
+ [11] vctrs_0.6.5 reshape2_1.4.4
+ [13] stringr_1.5.1 pkgconfig_2.0.3
+ [15] SpatialExperiment_1.14.0 crayon_1.5.3
+ [17] fastmap_1.2.0 backports_1.5.0
+ [19] magick_2.8.4 XVector_0.44.0
+ [21] labeling_0.4.3 utf8_1.2.4
+ [23] rmarkdown_2.27 UCSC.utils_1.0.0
+ [25] ragg_1.3.2 purrr_1.0.2
+ [27] xfun_0.46 beachmat_2.20.0
+ [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
+ [31] jsonlite_1.8.8 DelayedArray_0.30.1
+ [33] BiocParallel_1.38.0 terra_1.7-78
+ [35] irlba_2.3.5.1 parallel_4.4.1
+ [37] R6_2.5.1 stringi_1.8.4
+ [39] RColorBrewer_1.1-3 reticulate_1.38.0
+ [41] parallelly_1.37.1 GenomicRanges_1.56.1
+ [43] scattermore_1.2 Rcpp_1.0.13
+ [45] bookdown_0.40 SummarizedExperiment_1.34.0
+ [47] knitr_1.48 future.apply_1.11.2
+ [49] R.utils_2.12.3 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/reference-keys.txt b/reference-keys.txt
index a17604d..555f956 100644
--- a/reference-keys.txt
+++ b/reference-keys.txt
@@ -4,115 +4,116 @@ fig:unnamed-chunk-14
fig:unnamed-chunk-15
fig:unnamed-chunk-16
fig:unnamed-chunk-18
-fig:unnamed-chunk-30
-fig:unnamed-chunk-35
-fig:unnamed-chunk-38
-fig:unnamed-chunk-40
+fig:unnamed-chunk-32
+fig:unnamed-chunk-37
fig:unnamed-chunk-42
-fig:unnamed-chunk-44
+fig:unnamed-chunk-45
+fig:unnamed-chunk-47
fig:unnamed-chunk-49
fig:unnamed-chunk-51
-fig:unnamed-chunk-53
-fig:unnamed-chunk-55
-fig:unnamed-chunk-59
-fig:unnamed-chunk-61
-fig:unnamed-chunk-63
+fig:unnamed-chunk-56
+fig:unnamed-chunk-58
+fig:unnamed-chunk-60
+fig:unnamed-chunk-62
fig:unnamed-chunk-66
-fig:unnamed-chunk-69
-fig:unnamed-chunk-74
+fig:unnamed-chunk-68
+fig:unnamed-chunk-70
+fig:unnamed-chunk-73
fig:unnamed-chunk-76
-fig:unnamed-chunk-78
-fig:unnamed-chunk-80
-fig:unnamed-chunk-82
-fig:unnamed-chunk-84
+fig:unnamed-chunk-81
+fig:unnamed-chunk-83
+fig:unnamed-chunk-85
fig:unnamed-chunk-87
+fig:unnamed-chunk-89
+fig:unnamed-chunk-91
fig:unnamed-chunk-94
-fig:unnamed-chunk-96
-fig:unnamed-chunk-98
fig:unnamed-chunk-101
fig:unnamed-chunk-103
fig:unnamed-chunk-105
+fig:unnamed-chunk-108
+fig:unnamed-chunk-110
fig:unnamed-chunk-112
-fig:unnamed-chunk-114
-fig:unnamed-chunk-118
-fig:unnamed-chunk-120
-fig:unnamed-chunk-123
-fig:unnamed-chunk-128
-fig:unnamed-chunk-131
-fig:unnamed-chunk-134
-fig:unnamed-chunk-137
+fig:unnamed-chunk-119
+fig:unnamed-chunk-121
+fig:unnamed-chunk-125
+fig:unnamed-chunk-127
+fig:unnamed-chunk-130
+fig:unnamed-chunk-135
+fig:unnamed-chunk-138
fig:unnamed-chunk-141
-fig:unnamed-chunk-143
-fig:unnamed-chunk-147
+fig:unnamed-chunk-144
+fig:unnamed-chunk-148
fig:unnamed-chunk-150
-fig:unnamed-chunk-152
+fig:unnamed-chunk-154
+fig:unnamed-chunk-157
fig:unnamed-chunk-159
-fig:unnamed-chunk-161
-fig:unnamed-chunk-163
-fig:unnamed-chunk-165
-fig:unnamed-chunk-167
-fig:unnamed-chunk-169
-fig:unnamed-chunk-171
+fig:unnamed-chunk-166
+fig:unnamed-chunk-168
+fig:unnamed-chunk-170
+fig:unnamed-chunk-172
fig:unnamed-chunk-174
-fig:unnamed-chunk-175
fig:unnamed-chunk-176
-fig:unnamed-chunk-185
-fig:unnamed-chunk-191
-fig:unnamed-chunk-194
+fig:unnamed-chunk-178
+fig:unnamed-chunk-181
+fig:unnamed-chunk-182
+fig:unnamed-chunk-183
+fig:unnamed-chunk-192
+fig:unnamed-chunk-198
fig:unnamed-chunk-201
-fig:unnamed-chunk-204
-fig:unnamed-chunk-206
-fig:unnamed-chunk-210
-fig:unnamed-chunk-212
-fig:unnamed-chunk-215
+fig:unnamed-chunk-208
+fig:unnamed-chunk-211
+fig:unnamed-chunk-213
fig:unnamed-chunk-217
fig:unnamed-chunk-219
-fig:unnamed-chunk-221
+fig:unnamed-chunk-222
fig:unnamed-chunk-224
-fig:unnamed-chunk-225
-fig:unnamed-chunk-227
-fig:unnamed-chunk-229
-fig:unnamed-chunk-230
+fig:unnamed-chunk-226
+fig:unnamed-chunk-228
+fig:unnamed-chunk-231
fig:unnamed-chunk-232
fig:unnamed-chunk-234
fig:unnamed-chunk-236
-fig:unnamed-chunk-275
-fig:unnamed-chunk-279
-fig:unnamed-chunk-281
-fig:unnamed-chunk-283
+fig:unnamed-chunk-237
+fig:unnamed-chunk-239
+fig:unnamed-chunk-241
+fig:unnamed-chunk-243
+fig:unnamed-chunk-282
fig:unnamed-chunk-286
fig:unnamed-chunk-288
fig:unnamed-chunk-290
-fig:unnamed-chunk-292
-fig:unnamed-chunk-294
+fig:unnamed-chunk-293
+fig:unnamed-chunk-295
fig:unnamed-chunk-297
fig:unnamed-chunk-299
fig:unnamed-chunk-301
-fig:unnamed-chunk-303
-fig:unnamed-chunk-305
-fig:unnamed-chunk-365
-fig:unnamed-chunk-366
+fig:unnamed-chunk-304
+fig:unnamed-chunk-306
+fig:unnamed-chunk-308
+fig:unnamed-chunk-310
+fig:unnamed-chunk-312
fig:unnamed-chunk-372
+fig:unnamed-chunk-373
fig:unnamed-chunk-379
-fig:unnamed-chunk-381
-fig:unnamed-chunk-387
-fig:unnamed-chunk-389
-fig:unnamed-chunk-395
-fig:unnamed-chunk-397
-fig:unnamed-chunk-462
-fig:unnamed-chunk-465
+fig:unnamed-chunk-386
+fig:unnamed-chunk-388
+fig:unnamed-chunk-394
+fig:unnamed-chunk-396
+fig:unnamed-chunk-402
+fig:unnamed-chunk-404
fig:unnamed-chunk-469
-fig:unnamed-chunk-470
fig:unnamed-chunk-472
-fig:unnamed-chunk-474
fig:unnamed-chunk-476
-fig:unnamed-chunk-478
+fig:unnamed-chunk-477
fig:unnamed-chunk-479
fig:unnamed-chunk-481
+fig:unnamed-chunk-483
fig:unnamed-chunk-485
+fig:unnamed-chunk-486
fig:unnamed-chunk-488
-fig:unnamed-chunk-490
fig:unnamed-chunk-492
+fig:unnamed-chunk-495
+fig:unnamed-chunk-497
+fig:unnamed-chunk-499
giotto-suite-workshop-2024
instructors
topics-and-schedule
@@ -153,6 +154,10 @@ presentation-1
ecosystem
installation-python-environment
giotto-installation-1
+system-prerequisites
+installation-of-released-version
+installation-of-dev-branch-giotto-packages
+common-install-issues
python-environment
default-installation
custom-installs
diff --git a/search_index.json b/search_index.json
index 2736898..72ab7c3 100644
--- a/search_index.json
+++ b/search_index.json
@@ -1 +1 @@
-[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: Ensure your system To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak. pak::pak("drieslab/Giotto") 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
+[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: 4.3.1.1 System prerequisites for windows, Rtools needs to be installed a major dependency terra needs GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux 4.3.1.2 Installation of released version To install the currently released version of Giotto in a single step: pak::pak("drieslab/Giotto") This should automatically install all the Giotto dependencies and other Giotto module packages. 4.3.1.3 Installation of dev branch Giotto packages You can install dev branch versions by using devtools::install_github() Core module dev branchs: “drieslab/Giotto@suite_dev” “drieslab/GiottoVisuals@dev” “drieslab/GiottoClass@dev” “drieslab/GiottoUtils@dev” devtools::install_github("drieslab/GiottoClass@dev") pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages. 4.3.1.4 Common install issues If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work. devtools::install_version("Matrix", version = "1.6-5") If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it. install.packages("irlba", type = "source") ## # Downloading packages ------------------------------------------------------- ## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s] ## Successfully downloaded 1 package in 1.2 seconds. ## ## The following package(s) will be installed: ## - irlba [2.3.5.1] ## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu". ## ## # Installing packages -------------------------------------------------------- ## - Installing irlba ... OK [built from source and cached in 5.7s] ## Successfully installed 1 package in 5.7 seconds. 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are: python_path - when set, will activate a python environment save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render. save_plot - whether to save plots to the save_dir return_plot - whether to return the plot objects. When FALSE, only NULL is returned show_plot - whether to show the plot in the viewer These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function. library(Giotto) save_dir <- "results/01_session2/" # this call will also intialize the python env instrs <- createGiottoInstructions( save_dir = save_dir, # working directory is the default show_plot = FALSE, save_plot = TRUE, return_plot = FALSE, python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default ) force(instrs) Giotto object creation functions all have an instructions param for passing in instructions objects. giotto objects will also respond to the instructions() generic. test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object # example plot g <- GiottoData::loadGiottoMini("visium") instructions(g) <- instrs instructions(g, "show_plot") # instructions say not to plot to viewer spatPlot2D(g, show_image = TRUE, image_name = "image") # instead it will directly write to the results folder As an example, you can also set individual instructions instructions(g, "show_plot") <- TRUE spatPlot2D(g, show_image = TRUE, image_name = "image") Figure 4.2: example image output "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
diff --git a/spatial-multi-modal-analysis.html b/spatial-multi-modal-analysis.html
index 5bdab74..c78f952 100644
--- a/spatial-multi-modal-analysis.html
+++ b/spatial-multi-modal-analysis.html
@@ -563,48 +563,48 @@ 13.2.3 Spatial classes:# load in data
-library(Giotto)
-
-g <- GiottoData::loadGiottoMini("vizgen")
-
-activeSpatUnit(g) <- "aggregate"
-
-gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE)
-
-gimg <- getGiottoImage(g)
+
13.3 Examples of the simple transforms with a giottoPolygon
-rain <- rainbow(nrow(gpoly))
-
-line_width <- 0.3
-
-# par to setup the grid plotting layout
-p <- par(no.readonly = TRUE)
-par(mfrow=c(3,3))
-gpoly |>
- plot(main = "no transform", col = rain, lwd = line_width)
-
-gpoly |> spatShift(dx = 1000) |>
- plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
-
-gpoly |> spin(45) |>
- plot(main = "spin(45)", col = rain, lwd = line_width)
-
-gpoly |> rescale(fx = 10, fy = 6) |>
- plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
-
-gpoly |> flip(direction = "vertical") |>
- plot(main = "flip()", col = rain, lwd = line_width)
-
-gpoly |> t() |>
- plot(main = "t()", col = rain, lwd = line_width)
-
-gpoly |> shear(fx = 0.5) |>
- plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
-par(p)
+rain <- rainbow(nrow(gpoly))
+
+line_width <- 0.3
+
+# par to setup the grid plotting layout
+p <- par(no.readonly = TRUE)
+par(mfrow=c(3,3))
+gpoly |>
+ plot(main = "no transform", col = rain, lwd = line_width)
+
+gpoly |> spatShift(dx = 1000) |>
+ plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
+
+gpoly |> spin(45) |>
+ plot(main = "spin(45)", col = rain, lwd = line_width)
+
+gpoly |> rescale(fx = 10, fy = 6) |>
+ plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
+
+gpoly |> flip(direction = "vertical") |>
+ plot(main = "flip()", col = rain, lwd = line_width)
+
+gpoly |> t() |>
+ plot(main = "t()", col = rain, lwd = line_width)
+
+gpoly |> shear(fx = 0.5) |>
+ plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
+par(p)
@@ -617,20 +617,20 @@ 13.4 Affine transforms# create affine2d
-aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
+# create affine2d
+aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set.
-
-# append several simple transforms
-aff <- aff |>
- spatShift(dx = 1000) |>
- spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- flip(direction = "vertical") |>
- t() |>
- shear(fx = 0.5)
-force(aff)
+
+# append several simple transforms
+aff <- aff |>
+ spatShift(dx = 1000) |>
+ spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ flip(direction = "vertical") |>
+ t() |>
+ shear(fx = 0.5)
+force(aff)
<affine2d>
anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax)
rotate : -0.785398163397448 (rad)
@@ -639,35 +639,35 @@ 13.4 Affine transformsplot(aff)
+
We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests.
-
+
13.5 Image transforms
Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting.
spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image.
-# affine transform of image info matches with polygon info
-gimg |> affine(aff) |> plot()
-gpoly |> affine(aff) |>
- plot(add = TRUE,
- border = "cyan",
- lwd = 0.3)
+# affine transform of image info matches with polygon info
+gimg |> affine(aff) |> plot()
+gpoly |> affine(aff) |>
+ plot(add = TRUE,
+ border = "cyan",
+ lwd = 0.3)
-# affine of the giotto object
-g |> affine(aff) |>
- spatInSituPlotPoints(
- show_image = TRUE,
- feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
- feats_color_code = rainbow(4),
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- point_size = 0.1,
- use_overlap = FALSE
- )
+# affine of the giotto object
+g |> affine(aff) |>
+ spatInSituPlotPoints(
+ show_image = TRUE,
+ feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
+ feats_color_code = rainbow(4),
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ point_size = 0.1,
+ use_overlap = FALSE
+ )
Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used.
@@ -687,256 +687,256 @@ 13.6.1 Example dataset: Xenium Br
– cooresponding centroid locations
The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance.
-library(Giotto)
-instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
- save_plot = TRUE,
- show_plot = TRUE)
-options(timeout = 999999)
-download_dir <-file.path(getwd(),'/data/03_session2/')
-destfile <- file.path(download_dir,'Multimodal_registration.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
-Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
-Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
+library(Giotto)
+instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
+ save_plot = TRUE,
+ show_plot = TRUE)
+options(timeout = 999999)
+download_dir <-file.path(getwd(),'/data/03_session2/')
+destfile <- file.path(download_dir,'Multimodal_registration.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
+Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
+Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
13.6.2 Target Coordinate system
Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system.
-xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
-xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
-xen_cell_pl
+xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
+xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+xen_cell_pl
13.6.3 Visium to register
Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration
-G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
- gene_column_index = 2,
- png_name = 'tissue_hires_image.png',
- instructions = NULL)
-# In the meantime, calculate statistics for easier plot showing
-G_visium <- normalizeGiotto(G_visium)
-G_visium <- addStatistics(G_visium)
-V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
-V_origin
+G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
+ gene_column_index = 2,
+ png_name = 'tissue_hires_image.png',
+ instructions = NULL)
+# In the meantime, calculate statistics for easier plot showing
+G_visium <- normalizeGiotto(G_visium)
+G_visium <- addStatistics(G_visium)
+V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
+V_origin
The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform.
-# create affine2d
-aff <- affine(diag(c(1,1)))
-aff <- aff |>
- spin(90) |>
- flip(direction = "horizontal")
-force(aff)
-
-# Apply the transform
-V_tansformed <- affine(G_visium,aff)
-spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
-spatplot_to_register
+# create affine2d
+aff <- affine(diag(c(1,1)))
+aff <- aff |>
+ spin(90) |>
+ flip(direction = "horizontal")
+force(aff)
+
+# Apply the transform
+V_tansformed <- affine(G_visium,aff)
+spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
+spatplot_to_register
Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images.
Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app.
-
+
Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result.
-landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
-affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
-
-V_final <- affine(G_visium,affine_mtx %*% aff@affine)
-
-spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
-spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
+affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
+
+V_final <- affine(G_visium,affine_mtx %*% aff@affine)
+
+spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
+spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.3.1 Create Pseudo Visium dataset for comparison
Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance.
-V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
-spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
-spatPlot2D(V_final,
- show_network = T,
- network_color = 'blue',
- point_size = 1)
-center_to_center <- min(spat_network$distance)
-radius <- center_to_center*55/200
+V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
+spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
+spatPlot2D(V_final,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1)
+center_to_center <- min(spat_network$distance)
+radius <- center_to_center*55/200
Now we get the Pseudo Visium polygons
-Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
-stamp_dt <- circleVertices(radius = radius, npoints = 100)
-pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
-pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
-plot(pseudo_visium_poly)
+Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
+stamp_dt <- circleVertices(radius = radius, npoints = 100)
+pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
+pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
+plot(pseudo_visium_poly)
Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points
-xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
-gpoints <- createGiottoPoints(xen_transcripts)
-Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
- gpolygons = list('visium' = pseudo_visium_poly))
+xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
+gpoints <- createGiottoPoints(xen_transcripts)
+Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
+ gpolygons = list('visium' = pseudo_visium_poly))
Get gene expression information by overlapping polygon to points
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'visium')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "visium",
- feat_info = "rna",
- aggr_function = "sum")
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'visium')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "visium",
+ feat_info = "rna",
+ aggr_function = "sum")
Manipulate the expression for plotting
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'visium',
- name = 'plot')
-
-spatFeatPlot2D(Xen_obj,
- point_size = 3.5,
- expression_values = 'plot',
- show_image = F,
- feats = 'ERBB2')
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ name = 'plot')
+
+spatFeatPlot2D(Xen_obj,
+ point_size = 3.5,
+ expression_values = 'plot',
+ show_image = F,
+ feats = 'ERBB2')
Subset the registered Visium and plot same gene
-#get the extent of giotto points, xmin, xmax, ymin, ymax
-subset_extent <- ext(gpoints@spatVector)
-sub_visium <- subsetGiottoLocs(V_final,
- x_min = subset_extent[1],
- x_max = subset_extent[2],
- y_min = subset_extent[3],
- y_max = subset_extent[4])
-spatFeatPlot2D(sub_visium,
- point_size = 2,
- expression_values = 'scaled',
- show_image = F,
- feats = 'ERBB2')
+#get the extent of giotto points, xmin, xmax, ymin, ymax
+subset_extent <- ext(gpoints@spatVector)
+sub_visium <- subsetGiottoLocs(V_final,
+ x_min = subset_extent[1],
+ x_max = subset_extent[2],
+ y_min = subset_extent[3],
+ y_max = subset_extent[4])
+spatFeatPlot2D(sub_visium,
+ point_size = 2,
+ expression_values = 'scaled',
+ show_image = F,
+ feats = 'ERBB2')
13.6.4 Register post-Xenium H&E and IF image
For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics.
-HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
- imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
- micron = 0.2125)
-plot(HE_xen)
+HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
+ imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
+ micron = 0.2125)
+plot(HE_xen)
The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use.
-HE_xen <- HE_xen |> flip(direction = "vertical")
-gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
-plot(gimg_rast)
-#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
+HE_xen <- HE_xen |> flip(direction = "vertical")
+gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
+plot(gimg_rast)
+#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots
-gg <- ggplot2::ggplot()
-pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
-pl + ggplot2::geom_smooth() +
- ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+gg <- ggplot2::ggplot()
+pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
+pl + ggplot2::geom_smooth() +
+ ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.4.1 Add registered image information and compare RNA vs protein expression
With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF
-CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
-HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
-Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
- largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
+CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
+HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
+Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
+ largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
Get the cell polygons, as Xenium and IF are both subcellular resolution
-cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
-colnames(cellpoly_dt) <- c('poly_ID','x','y')
-cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
-Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
+cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
+colnames(cellpoly_dt) <- c('poly_ID','x','y')
+cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
+Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
Compute the gene expression matrix by overlay the cell polygons and giotto points.
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'cell')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "cell",
- feat_info = "rna",
- aggr_function = "sum")
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'rna',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'ERBB2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'cell')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "cell",
+ feat_info = "rna",
+ aggr_function = "sum")
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'rna',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'ERBB2',
+ point_size = 0.05)
Now we overlay the HER2 expression from the raster image with the cell polygons.
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'cell',
- image_names = c('HER2','CD20'))
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'HER2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'cell',
+ image_names = c('HER2','CD20'))
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'HER2',
+ point_size = 0.05)
We can also overlay the protein expression to Visium spots
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'visium',
- image_names = c('HER2','CD20'))
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "visium",
- feat_info = "protein",
- aggr_function = "sum")
-
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'visium',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'visium',
- feats = 'HER2',
- point_size = 2)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'visium',
+ image_names = c('HER2','CD20'))
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "visium",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'visium',
+ feats = 'HER2',
+ point_size = 2)
@@ -944,29 +944,29 @@ 13.6.4.1 Add registered image inf
13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation
Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially.
Here, we provide an example of two compressed images to show the automatic alignment pipeline.
-
+
-IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
-terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
+IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
+terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform.
-estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
- flip_horizontal = T,
- use_single_channel = T,
- single_channel_number = 3),
- y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
- invert = T),
- plot_match = T,
- max_ratio = 0.5,estimate_fun = 'Projective')
+estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
+ flip_horizontal = T,
+ use_single_channel = T,
+ single_channel_number = 3),
+ y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
+ invert = T),
+ plot_match = T,
+ max_ratio = 0.5,estimate_fun = 'Projective')
Use the estimation, we can quickly visualize the transformation
-mtx <- as.matrix(estimation$params)
-transformed <- affine(IF, mtx)
-To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
-plot(HE)
-plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
-SIFT_registration_overlay
+mtx <- as.matrix(estimation$params)
+transformed <- affine(IF, mtx)
+To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
+plot(HE)
+plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
+SIFT_registration_overlay
diff --git a/spatial-proteomics-multiplexed-immunofluorescence.html b/spatial-proteomics-multiplexed-immunofluorescence.html
index 684d4e8..f550962 100644
--- a/spatial-proteomics-multiplexed-immunofluorescence.html
+++ b/spatial-proteomics-multiplexed-immunofluorescence.html
@@ -518,33 +518,33 @@ 11 Spatial proteomics: Multiplexe
Junxiang Xu
August 6th 2024
Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so.
-#Install the Giotto Environment with Cellpose, note that we only need to do it once
-reticulate::conda_create(envname = "giotto_cellpose",
- python_version = 3.8)
-#.re.restartR()
-reticulate::use_condaenv('giotto_cellpose')
-reticulate::py_install(
- pip = TRUE,
- envname = 'giotto_cellpose',
- packages = c(
- "pandas",
- "networkx",
- "python-igraph",
- "leidenalg",
- "scikit-learn",
- "cellpose",
- "smfishhmrf",
- 'tifffile',
- 'scikit-image'
- )
-)
-#.rs.restartR()
+#Install the Giotto Environment with Cellpose, note that we only need to do it once
+reticulate::conda_create(envname = "giotto_cellpose",
+ python_version = 3.8)
+#.re.restartR()
+reticulate::use_condaenv('giotto_cellpose')
+reticulate::py_install(
+ pip = TRUE,
+ envname = 'giotto_cellpose',
+ packages = c(
+ "pandas",
+ "networkx",
+ "python-igraph",
+ "leidenalg",
+ "scikit-learn",
+ "cellpose",
+ "smfishhmrf",
+ 'tifffile',
+ 'scikit-image'
+ )
+)
+#.rs.restartR()
Now, activate the Giotto python environment.
-#.rs.restartR()
-# Activate the Giotto python environment of your choice
-GiottoClass::set_giotto_python_path('giotto_cellpose')
-# Check if cellpose was successfully installed
-GiottoUtils::package_check('cellpose',repository = 'pip')
+#.rs.restartR()
+# Activate the Giotto python environment of your choice
+GiottoClass::set_giotto_python_path('giotto_cellpose')
+# Check if cellpose was successfully installed
+GiottoUtils::package_check('cellpose',repository = 'pip')
11.1 Spatial Proteomics Technologies
This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix.
@@ -557,58 +557,58 @@
11.2 Raw data type coming out of
11.2.1 Use ome.tiff as an example output data to begin with
OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore.
-library(Giotto)
-instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
- save_plot = TRUE,
- show_plot = TRUE,
- python_path = 'giotto_cellpose')
-options(timeout = 9999999)
-download_dir =file.path(getwd(),'/data/02_session4/')
-destfile = file.path(download_dir,'Lunaphore.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
-list.files(paste0(download_dir,'/Lunaphore'))
+library(Giotto)
+instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ python_path = 'giotto_cellpose')
+options(timeout = 9999999)
+download_dir =file.path(getwd(),'/data/02_session4/')
+destfile = file.path(download_dir,'Lunaphore.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
+list.files(paste0(download_dir,'/Lunaphore'))
We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML.
-img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
-img_meta
+img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
+img_meta
However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it.
-##Get channel information
-Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
-node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
-channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
-channel_df
+##Get channel information
+Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
+node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
+channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
+channel_df
11.2.2 Use single channel images as an example output data to begin with
Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like.
-#Create multichannel raster and extract each single channels
-Luna_terra = terra::rast(Luna)
-names(Luna_terra) <- channel_df$Name
-gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
-plot(gimg_DAPI)
+#Create multichannel raster and extract each single channels
+Luna_terra = terra::rast(Luna)
+names(Luna_terra) <- channel_df$Name
+gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
+plot(gimg_DAPI)
Extract and save the raster image for future use.
-single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
-if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
-for (i in 1:nrow(channel_df)){
- single_channel <- terra::subset(Luna_terra, i)
- terra::writeRaster(single_channel,
- filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
- overwrite=T)
-}
+single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
+if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
+for (i in 1:nrow(channel_df)){
+ single_channel <- terra::subset(Luna_terra, i)
+ terra::writeRaster(single_channel,
+ filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
+ overwrite=T)
+}
Create a list of GiottoLargeImages using single channel rasters.
-file_names = list.files(single_channel_dir,full.names = TRUE)
-image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
-gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
- names = image_names,
- negative_y = F,
- flip_vertical = T)
-names(gimg_list) <- image_names
-plot(gimg_list[['Vimentin']])
+file_names = list.files(single_channel_dir,full.names = TRUE)
+image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
+gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
+ names = image_names,
+ negative_y = F,
+ flip_vertical = T)
+names(gimg_list) <- image_names
+plot(gimg_list[['Vimentin']])
@@ -618,34 +618,34 @@ 11.3 Cell Segmentation file to ge
11.3.1 Using segmentation output file from DeepCell(mesmer) as an example.
We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function.
-gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
-plot(gpoly_mesmer)
+gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
+plot(gpoly_mesmer)
We can also zoom in to check how does the segmentation look.
-zoom <- c(2000,2500,2000,2500)
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+zoom <- c(2000,2500,2000,2500)
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
11.3.2 Using Giotto wrapper of Cellpose to perform segmentation
-gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
-writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
-#Create a giotto image to evaluate segmentation
-gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
+gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
+writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
+#Create a giotto image to evaluate segmentation
+gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website
-doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
- mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- channel_1 = 0,
- channel_2 = 0,
- model_name = 'cyto3',
- batch_size=12)
-cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- shift_horizontal_step = F,
- shift_vertical_step = F,
- flip_vertical = T)
-plot(gimg_for_cellpose)
-plot(cpoly, add = T, border = 'red')
+doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
+ mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ channel_1 = 0,
+ channel_2 = 0,
+ model_name = 'cyto3',
+ batch_size=12)
+cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ shift_horizontal_step = F,
+ shift_vertical_step = F,
+ flip_vertical = T)
+plot(gimg_for_cellpose)
+plot(cpoly, add = T, border = 'red')
@@ -654,133 +654,133 @@ 11.4 Create a Giotto Object using
You will need to have
- list of giotto images
- giottoPolygon created from segmentation
-Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
- images = gimg_list,
- instructions = instrs)
-Lunaphore_giotto
+Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
+ images = gimg_list,
+ instructions = instrs)
+Lunaphore_giotto
11.4.1 Overlap to matrix
calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with
-Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
- spatial_info = 'cell',
- image_names = names(gimg_list))
-
-Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
- type ='intensity',
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-showGiottoExpression(Lunaphore_giotto)
+Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
+ spatial_info = 'cell',
+ image_names = names(gimg_list))
+
+Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
+ type ='intensity',
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+showGiottoExpression(Lunaphore_giotto)
11.4.2 Manipulate Expression information
For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile.
Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object
-expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
-filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
-Lunaphore_giotto <- setExpression(Lunaphore_giotto,
- feat_type = 'protein',
- x = createExprObj(filtered_expr_mtx),
- name = 'raw')
-showGiottoExpression(Lunaphore_giotto)
+expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
+filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
+Lunaphore_giotto <- setExpression(Lunaphore_giotto,
+ feat_type = 'protein',
+ x = createExprObj(filtered_expr_mtx),
+ name = 'raw')
+showGiottoExpression(Lunaphore_giotto)
11.4.3 Rescale polygons
rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis
-Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
- poly_info = 'cell',
- name = 'smallcell',
- fx = 0.7, fy = 0.7, calculate_centroids = T)
-smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
-plot(smallpoly,add = TRUE, border = "red",ext = zoom)
+Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
+ poly_info = 'cell',
+ name = 'smallcell',
+ fx = 0.7, fy = 0.7, calculate_centroids = T)
+smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+plot(smallpoly,add = TRUE, border = "red",ext = zoom)
11.4.4 Perform clustering and differential expression
The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering
-Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
-Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
-
-Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
- scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
- set_seed = TRUE)
-screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
+Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
+Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
+
+Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
+ scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
+ set_seed = TRUE)
+screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check.
-Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
- dimensions_to_use = 1:5,
- set_seed = TRUE)
-Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
-Lunaphore_giotto = doHclust(Lunaphore_giotto,
- k = 8,
- dim_reduction_to_use = 'cells',
- spat_unit = 'cell',
- feat_type = 'protein')
-
-spatInSituPlotPoints(gobject = Lunaphore_giotto,
- spat_unit = "cell",
- polygon_feat_type = "cell",
- show_polygon = T,
- feat_type = "protein",
- feats = NULL,
- polygon_fill = 'hclust',
- polygon_fill_as_factor = TRUE,
- polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
- polygon_color = 'black',
- background_color = "white")
+Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
+ dimensions_to_use = 1:5,
+ set_seed = TRUE)
+Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
+Lunaphore_giotto = doHclust(Lunaphore_giotto,
+ k = 8,
+ dim_reduction_to_use = 'cells',
+ spat_unit = 'cell',
+ feat_type = 'protein')
+
+spatInSituPlotPoints(gobject = Lunaphore_giotto,
+ spat_unit = "cell",
+ polygon_feat_type = "cell",
+ show_polygon = T,
+ feat_type = "protein",
+ feats = NULL,
+ polygon_fill = 'hclust',
+ polygon_fill_as_factor = TRUE,
+ polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
+ polygon_color = 'black',
+ background_color = "white")
Then we can check the heatmap of protein expression and determine the first round of cluster annotation
-cluster_column = 'hclust'
-plotMetaDataHeatmap(Lunaphore_giotto,
- spat_unit = 'cell',feat_type = 'protein',
- expression_values = "raw",
- metadata_cols = c(cluster_column),
- selected_feats = names(gimg_list),
- y_text_size = 8,
- show_values = 'zscores_rescaled')
+cluster_column = 'hclust'
+plotMetaDataHeatmap(Lunaphore_giotto,
+ spat_unit = 'cell',feat_type = 'protein',
+ expression_values = "raw",
+ metadata_cols = c(cluster_column),
+ selected_feats = names(gimg_list),
+ y_text_size = 8,
+ show_values = 'zscores_rescaled')
11.4.5 Give the cluster an annotation based on expression values
-
+
11.4.6 spatial network
This is to create a cellular neighborhood based on nearest neighbor of physical distance
-Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
-
-spatPlot2D(Lunaphore_giotto,
- show_network = T,
- network_color = 'blue',
- point_size = 1.5,
- cell_color = 'hclust')
+Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
+
+spatPlot2D(Lunaphore_giotto,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1.5,
+ cell_color = 'hclust')
11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions
This is using cellProximityEnrichment() to statistically identify cell type interactions
-cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
- cluster_column = 'cell_types',
- spatial_network_name = 'Delaunay_network',
- adjust_method = 'fdr',
- number_of_simulations = 2000)
-## barplot
-cellProximityBarplot(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- min_orig_ints = 5, min_sim_ints = 5)
+cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
+ cluster_column = 'cell_types',
+ spatial_network_name = 'Delaunay_network',
+ adjust_method = 'fdr',
+ number_of_simulations = 2000)
+## barplot
+cellProximityBarplot(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ min_orig_ints = 5, min_sim_ints = 5)
-## network
-cellProximityNetwork(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- remove_self_edges = T,
- only_show_enrichment_edges = F)
+## network
+cellProximityNetwork(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ remove_self_edges = T,
+ only_show_enrichment_edges = F)
diff --git a/visium-hd.html b/visium-hd.html
index 23aa210..f24cdd4 100644
--- a/visium-hd.html
+++ b/visium-hd.html
@@ -525,7 +525,7 @@ 9.1 Objective9.2 Background
9.2.1 Visium HD Technology
-
+
Figure 9.1: Overview of Visium HD. Source: 10X Genomics
@@ -536,7 +536,7 @@
9.2.1 Visium HD Technology
9.2.2 Colorectal Cancer Sample
-
+
Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics
@@ -550,7 +550,7 @@
9.2.2 Colorectal Cancer Sample9.3 Data Ingestion
9.3.1 Visium HD output data format
-
+
Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline.
@@ -560,19 +560,19 @@
9.3.1 Visium HD output data forma
9.4 Tiling and aggregation
The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells.
-
+
9.5 Scalability and projection functions
diff --git a/visium-part-i.html b/visium-part-i.html
index 5aa1658..c4746ee 100644
--- a/visium-part-i.html
+++ b/visium-part-i.html
@@ -520,7 +520,7 @@ 7 Visium Part I
7.1 The Visium technology
Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression.
-
+
Figure 7.1: Visum workflow. Source: 10X Genomics
@@ -536,73 +536,73 @@
7.2 Introduction to the spatial d
7.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/01_session5/")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
+dir.create("data/01_session5/")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”.
-
+
7.4 Create the Giotto object
createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/01_session5/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-## Provide the path to the visium folder
-data_path <- "data/01_session5/"
-
-## Create object directly from the visium folder
-visium_brain <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/01_session5/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+## Provide the path to the visium folder
+data_path <- "data/01_session5/"
+
+## Create object directly from the visium folder
+visium_brain <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
7.5 Subset on spots that were covered by tissue
Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area.
-spatPlot2D(
- gobject = visium_brain,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE)
-
+spatPlot2D(
+ gobject = visium_brain,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE)
+
Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area.
-
+
7.6 Quality control
@@ -610,43 +610,43 @@ 7.6 Quality controlvisium_brain_statistics <- addStatistics(gobject = visium_brain,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = visium_brain_statistics,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain_statistics <- addStatistics(gobject = visium_brain,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = visium_brain_statistics,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.3: Spatial distribution of features per spot.
filterDistributions() creates a histogram to show the distribution of features per spot across the sample.
-
-
+
+
Figure 7.4: Distribution of features per spot.
When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample.
-
-
+
+
Figure 7.5: Distribution of cells with different features per spot.
filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data:
-filterCombinations(gobject = visium_brain_statistics,
- expression_thresholds = c(1, 2, 3),
- feat_det_in_min_cells = c(50, 100, 200),
- min_det_feats_per_cell = c(500, 1000, 1500))
-
+filterCombinations(gobject = visium_brain_statistics,
+ expression_thresholds = c(1, 2, 3),
+ feat_det_in_min_cells = c(50, 100, 200),
+ min_det_feats_per_cell = c(500, 1000, 1500))
+
Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations.
@@ -656,34 +656,34 @@
7.6 Quality control
7.7 Filtering
Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample.
-visium_brain <- filterGiotto(
- gobject = visium_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE
-)
-Feature type: rna
-Number of cells removed: 4 out of 2702
-Number of feats removed: 7311 out of 22125
+
+
7.8 Normalization
Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one.
-visium_brain <- normalizeGiotto(
- gobject = visium_brain,
- scalefactor = 6000,
- verbose = TRUE
-)
+visium_brain <- normalizeGiotto(
+ gobject = visium_brain,
+ scalefactor = 6000,
+ verbose = TRUE
+)
Calculate the normalized number of features per spot and save the statistics in the metadata table.
-visium_brain <- addStatistics(gobject = visium_brain)
-
-## visualize
-spatPlot2D(gobject = visium_brain,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain <- addStatistics(gobject = visium_brain)
+
+## visualize
+spatPlot2D(gobject = visium_brain,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.7: Spatial distribution of the number of features per spot.
@@ -698,11 +698,11 @@
7.9.1 Highly Variable Features:
loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_loess",
- save_plot = TRUE,
- default_save_name = "HVFplot_loess")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_loess",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_loess")
+
Figure 7.8: Covariance of HVFs using the loess method.
@@ -711,11 +711,11 @@
7.9.1 Highly Variable Features:
pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")
+
Figure 7.9: Variance of HVFs using the pearson residuals method.
@@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
17.4.5 Check polygon information
We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots.
- -
+
+
-
Figure 17.7: Polygon information for stardist_cell. @@ -727,23 +727,23 @@
17.5 Performing kriging17.5.1 Interpolating features
Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required.
-future::plan(future::multisession()) # comment out for single threading
-subcell_v_brain <- interpolateFeature(v_brain,
- spat_unit = 'cell',
- feat_type = 'rna',
- ext = ext(v_brain),
- feats = ext_spatial_features,
- overwrite = T)
-
-print(subcell_v_brain)
+
+
future::plan(future::multisession()) # comment out for single threading
+subcell_v_brain <- interpolateFeature(v_brain,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ ext = ext(v_brain),
+ feats = ext_spatial_features,
+ overwrite = T)
+
+print(subcell_v_brain)Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right).
For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1.
-
+
+
After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified.
-
Figure 17.9: Raster of gene expression interpolation for Pantr1 @@ -754,18 +754,18 @@
17.5.1 Interpolating features17.5.2 Expression overlap
The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier.
This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run.
-subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain,
- name_overlap = "rna",
- spatial_info = "stardist_cell",
- image_names = ext_spatial_features)
-
-subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain,
- poly_info = "stardist_cell",
- feat_info = "rna",
- aggr_function = "sum",
- type='intensity')subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain,
+ name_overlap = "rna",
+ spatial_info = "stardist_cell",
+ image_names = ext_spatial_features)
+
+subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain,
+ poly_info = "stardist_cell",
+ feat_info = "rna",
+ aggr_function = "sum",
+ type='intensity')
+
17.5.2 Expression overlap
+
+
Figure 17.10: Gene expression for cells based on interpolation. @@ -776,70 +776,70 @@
17.5.2 Expression overlap
17.6 Reading in larger dataset
For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here.
-
+
17.7 Analyzing interpolated features
17.7.1 Filter and normalization
Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis.
-subcell_v_brain <- filterGiotto(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- expression_values = "raw",
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-
-subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- scalefactor = 6000,
- verbose = T)subcell_v_brain <- filterGiotto(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ expression_values = "raw",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+
+subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ scalefactor = 6000,
+ verbose = T)17.7.2 Visualizing gene expression from interpolated expression
Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data.
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Sparc',
- point_size = 2,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_sparc_vis'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'stardist_cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Sparc',
- point_size = 0.6,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_sparc'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'cell',
- gradient_style = 'sequential',
- feats = 'Pantr1',
- cell_color_gradient = 'Geyser',
- point_size = 2,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_pantr1_vis'))
-
-spatFeatPlot2D(subcell_v_brain,
- spat_unit = 'stardist_cell',
- gradient_style = 'sequential',
- cell_color_gradient = 'Geyser',
- feats = 'Pantr1',
- point_size = 0.6,
- save_plot = TRUE,
- show_image = TRUE,
- save_param = list(save_name = '03_ses6_pantr1'))spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Sparc',
+ point_size = 2,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_sparc_vis'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Sparc',
+ point_size = 0.6,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_sparc'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'cell',
+ gradient_style = 'sequential',
+ feats = 'Pantr1',
+ cell_color_gradient = 'Geyser',
+ point_size = 2,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_pantr1_vis'))
+
+spatFeatPlot2D(subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ gradient_style = 'sequential',
+ cell_color_gradient = 'Geyser',
+ feats = 'Pantr1',
+ point_size = 0.6,
+ save_plot = TRUE,
+ show_image = TRUE,
+ save_param = list(save_name = '03_ses6_pantr1'))Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer.
-
+
-
+
+
diff --git a/multi-omics-integration.html b/multi-omics-integration.html
index 47cad23..e76e25a 100644
--- a/multi-omics-integration.html
+++ b/multi-omics-integration.html
@@ -520,14 +520,14 @@ 14 Multi-omics integration
Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). @@ -848,37 +848,37 @@
17.7.2 Visualizing gene expressio
17.7.3 Run PCA
- +17.7.4 Clustering
-# UMAP
-subcell_v_brain <- runUMAP(subcell_v_brain,
- spat_unit = "stardist_cell",
- dimensions_to_use = 1:15,
- n_neighbors = 1000,
- min_dist = 0.001,
- spread = 1)
-
-# NN Network
-subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- dimensions_to_use = 1:10,
- feats_to_use = hv_feats,
- expression_values = 'normalized',
- k = 70)
-
-subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain,
- spat_unit = "stardist_cell",
- resolution = 0.15,
- n_iterations = 100,
- partition_type = 'RBConfigurationVertexPartition')
-
-plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus')
+
+
17.7.4 Clustering
# UMAP
+subcell_v_brain <- runUMAP(subcell_v_brain,
+ spat_unit = "stardist_cell",
+ dimensions_to_use = 1:15,
+ n_neighbors = 1000,
+ min_dist = 0.001,
+ spread = 1)
+
+# NN Network
+subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ dimensions_to_use = 1:10,
+ feats_to_use = hv_feats,
+ expression_values = 'normalized',
+ k = 70)
+
+subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain,
+ spat_unit = "stardist_cell",
+ resolution = 0.15,
+ n_iterations = 100,
+ partition_type = 'RBConfigurationVertexPartition')
+
+plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus')Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. @@ -888,27 +888,27 @@
17.7.4 Clustering
17.7.5 Visualizing clustering
Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster.
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.5,
- point_shape = 'no_border',
- background_color = 'black',
- save_plot = F,
- show_legend = TRUE)
-
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat'))
-
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.5,
+ point_shape = 'no_border',
+ background_color = 'black',
+ save_plot = F,
+ show_legend = TRUE)
+
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat'))
+
Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
@@ -918,37 +918,37 @@
17.7.5 Visualizing clustering
17.7.6 Cropping objects
We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below.
-subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
- spat_unit = ":all:",
- x_min = 4000,
- x_max = 7000,
- y_min = -6500,
- y_max = -3500,
- z_max = NULL,
- z_min = NULL)
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 2,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_vis_spat_crop'))
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat_crop'))
-
+subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
+ spat_unit = ":all:",
+ x_min = 4000,
+ x_max = 7000,
+ y_min = -6500,
+ y_max = -3500,
+ z_max = NULL,
+ z_min = NULL)
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 2,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_vis_spat_crop'))
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat_crop'))
+
Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
diff --git a/interoperability-with-isolated-tools.html b/interoperability-with-isolated-tools.html
index ceb2d0a..8cca7e6 100644
--- a/interoperability-with-isolated-tools.html
+++ b/interoperability-with-isolated-tools.html
@@ -526,1521 +526,1521 @@
16.1.1 Dataset downloaddata_path <- "data"
-dir.create(data_path)
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
- destfile = file.path(data_path,"counts.h5ad"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
- destfile = file.path(data_path,"cell_labels.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
+data_path <- "data"
+dir.create(data_path)
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
+ destfile = file.path(data_path,"counts.h5ad"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
+ destfile = file.path(data_path,"cell_labels.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
16.1.2 Create the Giotto object
-library(Giotto)
-library(reticulate)
-
-## Set instructions
-results_folder <- "results"
-dir.create(results_folder)
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-
-## load data
-
-### meta data
-meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-colnames(meta_df)[[1]] <- "cell_ID"
-slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
-slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
-
-### counts matrix
-scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
-ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
-reticulate::source_python(ad2g_path)
-adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
-X <- extract_expression(adata)
-cID <- extract_cell_IDs(adata)
-fID <- extract_feat_IDs(adata)
-rownames(X) <- fID
-colnames(X) <- cID
-slice1_filter <- cID %in% slice1_meta_df$cell_ID
-slice2_filter <- cID %in% slice2_meta_df$cell_ID
-slice1_exp <- X[,slice1_filter]
-slice2_exp <- X[,slice2_filter]
-
-### cell segmentation to cell location
-segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_df <- rbind(segments_1_df, segments_2_df)
-loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
-loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
-loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
-centr.x <- apply(loc.use[,loc.x],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-centr.y <- apply(loc.use[,loc.y],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
-slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
-slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
-
-## giotto obj
-giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
- spatial_locs = slice1_spatial_locs_df,
- cell_metadata = slice1_meta_df)
-giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
- spatial_locs = slice2_spatial_locs_df,
- cell_metadata = slice2_meta_df)
-giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
- gobject_names = c('mouse2_slice229',
- 'mouse2_slice300'), # name for each samples
- join_method = 'z_stack')
-# We skip the processing process here to save time and use the given cell type
-# annotation directly
-
-ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
- cell_type = "subclass",
- output_path = data_path,
- spat_unit = "cell",
- feat_type = "rna",
- verbose = TRUE)
-
-# Cell_ID Sample x y Cell_Type
-# <char> <char> <dbl> <dbl> <char>
-# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
-# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
-# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
-# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
-# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
-# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
+library(Giotto)
+library(reticulate)
+
+## Set instructions
+results_folder <- "results"
+dir.create(results_folder)
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+
+## load data
+
+### meta data
+meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+colnames(meta_df)[[1]] <- "cell_ID"
+slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
+slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
+
+### counts matrix
+scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
+ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
+reticulate::source_python(ad2g_path)
+adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
+X <- extract_expression(adata)
+cID <- extract_cell_IDs(adata)
+fID <- extract_feat_IDs(adata)
+rownames(X) <- fID
+colnames(X) <- cID
+slice1_filter <- cID %in% slice1_meta_df$cell_ID
+slice2_filter <- cID %in% slice2_meta_df$cell_ID
+slice1_exp <- X[,slice1_filter]
+slice2_exp <- X[,slice2_filter]
+
+### cell segmentation to cell location
+segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_df <- rbind(segments_1_df, segments_2_df)
+loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
+loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
+loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
+centr.x <- apply(loc.use[,loc.x],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+centr.y <- apply(loc.use[,loc.y],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
+slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
+slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
+
+## giotto obj
+giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
+ spatial_locs = slice1_spatial_locs_df,
+ cell_metadata = slice1_meta_df)
+giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
+ spatial_locs = slice2_spatial_locs_df,
+ cell_metadata = slice2_meta_df)
+giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
+ gobject_names = c('mouse2_slice229',
+ 'mouse2_slice300'), # name for each samples
+ join_method = 'z_stack')
+# We skip the processing process here to save time and use the given cell type
+# annotation directly
+
+ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
+ cell_type = "subclass",
+ output_path = data_path,
+ spat_unit = "cell",
+ feat_type = "rna",
+ verbose = TRUE)
+
+# Cell_ID Sample x y Cell_Type
+# <char> <char> <dbl> <dbl> <char>
+# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
+# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
+# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
+# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
+# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
+# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
Spatial cell type distributions
-spatPlot2D(giotto_obj,
- group_by = "slice_id",
- cell_color = "subclass",
- point_size = 1,
- point_border_stroke = NA,
- legend_text = 6)
+spatPlot2D(giotto_obj,
+ group_by = "slice_id",
+ cell_color = "subclass",
+ point_size = 1,
+ point_border_stroke = NA,
+ legend_text = 6)
16.1.2.1 Installation ONTraC
You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node.
It will run for less than 10 minutes on this example dataset.
For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time.
-source ~/.bash_profile
-conda create -y -n ONTraC python=3.11
-conda activate ONTraC
-pip install git+https://github.com/gyuanlab/ONTraC.git@V2
-# Retrieving notices: ...working... done
-# Channels:
-# - conda-forge
-# - bioconda
-# - r
-# - pytorch
-# - defaults
-# Platform: osx-arm64
-# Collecting package metadata (repodata.json): ...working... done
-# Solving environment: ...working... done
-#
-# ## Package Plan ##
-#
-# environment location: /Users/wwang/miniconda3/envs/ONTraC
-#
-# added / updated specs:
-# - python=3.11
-#
-#
-# The following packages will be downloaded:
-#
-# package | build
-# ---------------------------|-----------------
-# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
-# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
-# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
-# ------------------------------------------------------------
-# Total: 4.3 MB
-#
-# The following NEW packages will be INSTALLED:
-#
-# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
-# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
-# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
-# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
-# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
-# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
-# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
-# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
-# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
-# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
-# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
-# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
-# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
-# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
-# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
-# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
-#
-#
-#
-# Downloading and Extracting Packages: ...working... done
-# Preparing transaction: ...working... done
-# Verifying transaction: ...working... done
-# Executing transaction: ...working... done
-# #
-# # To activate this environment, use
-# #
-# # $ conda activate ONTraC
-# #
-# # To deactivate an active environment, use
-# #
-# # $ conda deactivate
-#
-# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
-# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
-# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
-# Running command git checkout -b V2 --track origin/V2
-# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
-# Installing build dependencies: started
-# Switched to a new branch 'V2'
-# branch 'V2' set up to track 'origin/V2'.
-# Installing build dependencies: finished with status 'done'
-# Getting requirements to build wheel: started
-# Getting requirements to build wheel: finished with status 'done'
-# Preparing metadata (pyproject.toml): started
-# Preparing metadata (pyproject.toml): finished with status 'done'
-# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
-# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
-# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
-# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
-# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
-# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
-# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
-# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
-# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
-# Collecting session-info (from ONTraC==2.0a9.dev7)
-# Using cached session_info-1.0.0-py3-none-any.whl
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
-# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
-# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
-# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
-# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
-# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
-# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
-# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
-# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
-# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
-# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
-# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
-# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
-# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
-# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
-# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
-# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
-# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
-# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
-# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
-# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
-# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
-# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
-# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
-# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
-# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
-# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
-# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
-# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
-# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
-# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
-# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
-# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
-# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
-# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
-# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
-# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
-# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
-# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
-# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
-# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
-# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
-# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
-# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
-# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
-# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
-# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
-# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
-# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
-# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
-# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
-# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
-# Using cached idna-3.7-py3-none-any.whl (66 kB)
-# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
-# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
-# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
-# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
-# Building wheels for collected packages: ONTraC
-# Building wheel for ONTraC (pyproject.toml): started
-# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
-# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
-# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
-# Successfully built ONTraC
-# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
-# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
+source ~/.bash_profile
+conda create -y -n ONTraC python=3.11
+conda activate ONTraC
+pip install git+https://github.com/gyuanlab/ONTraC.git@V2
+# Retrieving notices: ...working... done
+# Channels:
+# - conda-forge
+# - bioconda
+# - r
+# - pytorch
+# - defaults
+# Platform: osx-arm64
+# Collecting package metadata (repodata.json): ...working... done
+# Solving environment: ...working... done
+#
+# ## Package Plan ##
+#
+# environment location: /Users/wwang/miniconda3/envs/ONTraC
+#
+# added / updated specs:
+# - python=3.11
+#
+#
+# The following packages will be downloaded:
+#
+# package | build
+# ---------------------------|-----------------
+# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
+# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
+# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
+# ------------------------------------------------------------
+# Total: 4.3 MB
+#
+# The following NEW packages will be INSTALLED:
+#
+# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
+# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
+# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
+# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
+# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
+# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
+# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
+# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
+# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
+# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
+# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
+# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
+# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
+# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
+# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
+# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
+#
+#
+#
+# Downloading and Extracting Packages: ...working... done
+# Preparing transaction: ...working... done
+# Verifying transaction: ...working... done
+# Executing transaction: ...working... done
+# #
+# # To activate this environment, use
+# #
+# # $ conda activate ONTraC
+# #
+# # To deactivate an active environment, use
+# #
+# # $ conda deactivate
+#
+# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
+# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
+# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
+# Running command git checkout -b V2 --track origin/V2
+# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
+# Installing build dependencies: started
+# Switched to a new branch 'V2'
+# branch 'V2' set up to track 'origin/V2'.
+# Installing build dependencies: finished with status 'done'
+# Getting requirements to build wheel: started
+# Getting requirements to build wheel: finished with status 'done'
+# Preparing metadata (pyproject.toml): started
+# Preparing metadata (pyproject.toml): finished with status 'done'
+# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
+# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
+# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
+# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
+# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
+# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
+# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
+# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
+# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
+# Collecting session-info (from ONTraC==2.0a9.dev7)
+# Using cached session_info-1.0.0-py3-none-any.whl
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
+# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
+# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
+# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
+# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
+# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
+# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
+# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
+# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
+# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
+# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
+# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
+# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
+# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
+# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
+# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
+# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
+# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
+# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
+# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
+# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
+# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
+# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
+# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
+# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
+# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
+# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
+# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
+# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
+# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
+# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
+# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
+# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
+# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
+# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
+# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
+# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
+# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
+# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
+# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
+# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
+# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
+# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
+# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
+# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
+# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
+# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
+# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
+# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
+# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
+# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
+# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
+# Using cached idna-3.7-py3-none-any.whl (66 kB)
+# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
+# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
+# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
+# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
+# Building wheels for collected packages: ONTraC
+# Building wheel for ONTraC (pyproject.toml): started
+# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
+# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
+# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
+# Successfully built ONTraC
+# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
+# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
16.1.2.2 Running ONTraC
-
-source ~/.bash_profile
-conda activate ONTraC_v2_py311
-ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
-# ##################################################################################
-#
-# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
-# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
-# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
-# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
-# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
-#
-# version: 2.0a11
-#
-# ##################################################################################
-# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
-# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
-# 11:33:23 --- INFO: -------- I/O options -------
-# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
-# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
-# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
-# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
-# 11:33:23 --- INFO: -------- preprocessing options -------
-# 11:33:23 --- INFO: resolution: 10.0
-# 11:33:23 --- INFO: -------- niche net constr options -------
-# 11:33:23 --- INFO: n_cpu: 4
-# 11:33:23 --- INFO: n_neighbors: 50
-# 11:33:23 --- INFO: n_local: 20
-# 11:33:23 --- INFO: embedding_adjust: False
-# 11:33:23 --- INFO: sigma: 1
-# 11:33:23 --- INFO: -------- train options -------
-# 11:33:23 --- INFO: device: cpu
-# 11:33:23 --- INFO: epochs: 1000
-# 11:33:23 --- INFO: batch_size: 0
-# 11:33:23 --- INFO: patience: 100
-# 11:33:23 --- INFO: min_delta: 0.001
-# 11:33:23 --- INFO: min_epochs: 50
-# 11:33:23 --- INFO: seed: 42
-# 11:33:23 --- INFO: lr: 0.03
-# 11:33:23 --- INFO: hidden_feats: 4
-# 11:33:23 --- INFO: k: 6
-# 11:33:23 --- INFO: modularity_loss_weight: 1.0
-# 11:33:23 --- INFO: purity_loss_weight: 100.0
-# 11:33:23 --- INFO: regularization_loss_weight: 0.1
-# 11:33:23 --- INFO: beta: 0.3
-# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
-# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
-# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
-# 11:33:23 --- INFO: ------------- Niche network construct ---------------
-# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
-# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
-# 11:33:28 --- INFO: Generating samples.yaml file.
-# 11:33:28 --- INFO: ------------ Niche network construct end ------------
-# 11:33:28 --- INFO: ------------------------ GNN ------------------------
-# 11:33:28 --- INFO: Loading dataset.
-# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
-# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
-# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
-# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
-# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
-# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
-# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
-# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
-# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
-# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
-# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
-# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
-# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
-# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
-# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
-# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
-# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
-# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
-# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
-# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
-# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
-# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
-# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
-# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
-# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
-# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
-# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
-# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
-# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
-# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
-# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
-# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
-# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
-# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
-# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
-# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
-# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
-# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
-# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
-# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
-# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
-# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
-# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
-# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
-# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
-# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
-# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
-# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
-# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
-# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
-# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
-# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
-# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
-# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
-# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
-# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
-# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
-# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
-# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
-# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
-# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
-# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
-# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
-# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
-# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
-# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
-# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
-# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
-# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
-# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
-# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
-# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
-# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
-# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
-# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
-# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
-# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
-# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
-# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
-# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
-# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
-# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
-# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
-# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
-# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
-# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
-# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
-# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
-# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
-# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
-# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
-# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
-# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
-# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
-# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
-# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
-# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
-# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
-# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
-# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
-# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
-# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
-# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
-# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
-# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
-# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
-# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
-# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
-# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
-# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
-# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
-# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
-# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
-# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
-# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
-# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
-# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
-# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
-# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
-# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
-# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
-# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
-# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
-# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
-# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
-# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
-# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
-# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
-# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
-# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
-# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
-# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
-# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
-# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
-# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
-# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
-# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
-# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
-# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
-# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
-# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
-# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
-# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
-# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
-# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
-# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
-# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
-# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
-# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
-# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
-# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
-# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
-# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
-# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
-# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
-# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
-# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
-# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
-# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
-# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
-# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
-# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
-# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
-# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
-# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
-# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
-# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
-# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
-# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
-# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
-# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
-# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
-# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
-# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
-# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
-# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
-# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
-# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
-# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
-# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
-# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
-# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
-# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
-# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
-# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
-# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
-# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
-# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
-# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
-# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
-# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
-# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
-# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
-# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
-# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
-# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
-# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
-# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
-# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
-# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
-# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
-# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
-# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
-# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
-# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
-# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
-# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
-# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
-# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
-# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
-# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
-# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
-# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
-# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
-# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
-# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
-# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
-# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
-# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
-# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
-# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
-# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
-# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
-# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
-# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
-# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
-# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
-# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
-# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
-# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
-# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
-# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
-# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
-# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
-# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
-# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
-# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
-# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
-# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
-# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
-# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
-# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
-# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
-# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
-# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
-# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
-# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
-# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
-# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
-# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
-# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
-# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
-# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
-# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
-# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
-# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
-# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
-# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
-# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
-# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
-# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
-# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
-# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
-# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
-# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
-# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
-# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
-# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
-# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
-# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
-# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
-# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
-# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
-# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
-# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
-# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
-# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
-# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
-# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
-# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
-# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
-# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
-# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
-# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
-# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
-# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
-# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
-# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
-# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
-# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
-# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
-# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
-# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
-# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
-# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
-# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
-# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
-# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
-# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
-# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
-# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
-# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
-# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
-# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
-# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
-# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
-# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
-# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
-# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
-# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
-# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
-# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
-# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
-# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
-# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
-# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
-# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
-# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
-# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
-# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
-# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
-# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
-# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
-# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
-# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
-# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
-# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
-# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
-# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
-# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
-# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
-# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
-# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
-# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
-# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
-# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
-# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
-# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
-# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
-# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
-# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
-# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
-# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
-# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
-# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
-# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
-# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
-# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
-# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
-# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
-# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
-# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
-# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
-# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
-# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
-# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
-# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
-# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
-# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
-# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
-# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
-# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
-# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
-# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
-# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
-# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
-# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
-# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
-# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
-# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
-# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
-# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
-# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
-# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
-# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
-# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
-# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
-# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
-# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
-# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
-# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
-# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
-# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
-# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
-# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
-# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
-# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
-# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
-# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
-# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
-# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
-# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
-# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
-# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
-# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
-# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
-# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
-# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
-# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
-# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
-# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
-# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
-# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
-# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
-# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
-# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
-# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
-# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
-# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
-# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
-# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
-# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
-# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
-# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
-# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
-# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
-# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
-# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
-# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
-# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
-# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
-# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
-# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
-# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
-# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
-# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
-# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
-# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
-# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
-# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
-# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
-# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
-# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
-# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
-# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
-# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
-# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
-# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
-# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
-# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
-# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
-# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
-# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
-# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
-# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
-# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
-# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
-# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
-# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
-# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
-# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
-# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
-# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
-# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
-# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
-# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
-# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
-# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
-# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
-# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
-# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
-# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
-# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
-# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
-# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
-# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
-# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
-# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
-# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
-# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
-# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
-# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
-# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
-# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
-# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
-# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
-# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
-# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
-# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
-# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
-# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
-# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
-# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
-# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
-# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
-# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
-# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
-# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
-# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
-# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
-# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
-# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
-# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
-# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
-# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
-# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
-# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
-# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
-# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
-# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
-# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
-# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
-# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
-# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
-# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
-# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
-# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
-# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
-# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
-# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
-# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
-# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
-# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
-# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
-# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
-# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
-# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
-# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
-# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
-# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
-# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
-# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
-# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
-# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
-# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
-# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
-# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
-# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
-# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
-# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
-# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
-# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
-# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
-# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
-# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
-# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
-# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
-# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
-# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
-# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
-# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
-# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
-# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
-# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
-# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
-# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
-# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
-# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
-# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
-# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
-# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
-# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
-# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
-# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
-# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
-# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
-# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
-# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
-# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
-# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
-# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
-# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
-# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
-# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
-# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
-# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
-# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
-# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
-# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
-# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
-# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
-# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
-# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
-# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
-# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
-# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
-# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
-# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
-# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
-# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
-# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
-# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
-# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
-# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
-# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
-# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
-# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
-# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
-# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
-# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
-# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
-# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
-# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
-# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
-# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
-# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
-# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
-# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
-# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
-# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
-# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
-# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
-# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
-# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
-# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
-# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
-# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
-# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
-# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
-# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
-# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
-# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
-# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
-# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
-# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
-# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
-# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
-# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
-# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
-# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
-# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
-# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
-# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
-# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
-# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
-# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
-# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
-# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
-# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
-# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
-# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
-# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
-# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
-# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
-# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
-# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
-# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
-# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
-# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
-# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
-# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
-# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
-# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
-# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
-# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
-# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
-# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
-# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
-# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
-# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
-# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
-# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
-# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
-# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
-# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
-# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
-# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
-# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
-# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
-# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
-# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
-# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
-# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
-# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
-# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
-# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
-# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
-# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
-# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
-# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
-# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
-# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
-# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
-# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
-# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
-# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
-# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
-# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
-# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
-# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
-# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
-# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
-# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
-# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
-# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
-# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
-# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
-# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
-# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
-# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
-# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
-# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
-# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
-# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
-# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
-# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
-# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
-# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
-# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
-# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
-# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
-# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
-# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
-# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
-# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
-# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
-# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
-# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
-# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
-# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
-# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
-# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
-# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
-# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
-# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
-# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
-# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
-# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
-# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
-# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
-# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
-# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
-# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
-# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
-# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
-# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
-# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
-# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
-# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
-# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
-# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
-# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
-# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
-# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
-# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
-# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
-# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
-# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
-# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
-# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
-# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
-# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
-# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
-# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
-# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
-# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
-# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
-# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
-# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
-# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
-# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
-# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
-# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
-# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
-# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
-# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
-# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
-# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
-# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
-# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
-# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
-# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
-# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
-# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
-# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
-# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
-# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
-# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
-# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
-# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
-# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
-# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
-# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
-# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
-# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
-# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
-# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
-# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
-# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
-# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
-# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
-# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
-# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
-# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
-# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
-# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
-# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
-# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
-# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
-# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
-# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
-# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
-# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
-# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
-# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
-# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
-# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
-# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
-# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
-# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
-# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
-# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
-# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
-# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
-# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
-# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
-# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
-# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
-# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
-# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
-# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
-# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
-# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
-# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
-# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
-# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
-# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
-# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
-# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
-# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
-# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
-# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
-# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
-# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
-# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
-# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
-# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
-# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
-# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
-# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
-# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
-# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
-# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
-# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
-# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
-# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
-# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
-# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
-# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
-# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
-# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
-# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
-# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
-# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
-# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
-# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
-# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
-# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
-# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
-# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
-# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
-# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
-# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
-# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
-# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
-# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
-# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
-# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
-# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
-# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
-# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
-# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
-# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
-# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
-# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
-# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
-# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
-# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
-# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
-# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
-# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
-# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
-# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
-# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
-# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
-# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
-# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
-# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
-# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
-# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
-# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
-# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
-# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
-# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
-# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
-# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
-# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
-# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
-# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
-# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
-# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
-# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
-# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
-# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
-# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
-# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
-# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
-# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
-# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
-# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
-# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
-# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
-# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
-# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
-# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
-# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
-# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
-# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
-# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
-# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
-# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
-# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
-# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
-# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
-# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
-# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
-# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
-# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
-# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
-# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
-# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
-# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
-# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
-# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
-# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
-# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
-# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
-# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
-# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
-# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
-# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
-# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
-# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
-# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
-# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
-# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
-# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
-# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
-# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
-# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
-# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
-# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
-# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
-# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
-# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
-# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
-# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
-# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
-# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
-# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
-# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
-# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
-# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
-# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
-# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
-# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
-# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
-# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
-# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
-# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
-# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
-# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
-# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
-# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
-# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
-# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
-# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
-# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
-# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
-# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
-# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
-# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
-# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
-# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
-# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
-# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
-# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
-# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
-# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
-# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
-# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
-# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
-# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
-# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
-# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
-# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
-# 11:38:37 --- INFO: Training process end.
-# 11:38:37 --- INFO: Evaluating process start.
-# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
-# 11:38:37 --- INFO: Evaluating process end.
-# 11:38:37 --- INFO: Predicting process start.
-# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
-# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
-# 11:38:38 --- INFO: Predicting process end.
-# 11:38:38 --- INFO: --------------------- GNN end ----------------------
-# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
-# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
-# 11:38:38 --- INFO: Loading dataset.
-# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:38:39 --- INFO: Calculating NTScore for each niche.
-# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
-# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
-# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
-# 11:38:39 --- INFO: Output NTScore tables.
-# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
+
+source ~/.bash_profile
+conda activate ONTraC_v2_py311
+ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
+# ##################################################################################
+#
+# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
+# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
+# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
+# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
+# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
+#
+# version: 2.0a11
+#
+# ##################################################################################
+# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
+# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
+# 11:33:23 --- INFO: -------- I/O options -------
+# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
+# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
+# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
+# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
+# 11:33:23 --- INFO: -------- preprocessing options -------
+# 11:33:23 --- INFO: resolution: 10.0
+# 11:33:23 --- INFO: -------- niche net constr options -------
+# 11:33:23 --- INFO: n_cpu: 4
+# 11:33:23 --- INFO: n_neighbors: 50
+# 11:33:23 --- INFO: n_local: 20
+# 11:33:23 --- INFO: embedding_adjust: False
+# 11:33:23 --- INFO: sigma: 1
+# 11:33:23 --- INFO: -------- train options -------
+# 11:33:23 --- INFO: device: cpu
+# 11:33:23 --- INFO: epochs: 1000
+# 11:33:23 --- INFO: batch_size: 0
+# 11:33:23 --- INFO: patience: 100
+# 11:33:23 --- INFO: min_delta: 0.001
+# 11:33:23 --- INFO: min_epochs: 50
+# 11:33:23 --- INFO: seed: 42
+# 11:33:23 --- INFO: lr: 0.03
+# 11:33:23 --- INFO: hidden_feats: 4
+# 11:33:23 --- INFO: k: 6
+# 11:33:23 --- INFO: modularity_loss_weight: 1.0
+# 11:33:23 --- INFO: purity_loss_weight: 100.0
+# 11:33:23 --- INFO: regularization_loss_weight: 0.1
+# 11:33:23 --- INFO: beta: 0.3
+# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
+# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
+# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
+# 11:33:23 --- INFO: ------------- Niche network construct ---------------
+# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
+# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
+# 11:33:28 --- INFO: Generating samples.yaml file.
+# 11:33:28 --- INFO: ------------ Niche network construct end ------------
+# 11:33:28 --- INFO: ------------------------ GNN ------------------------
+# 11:33:28 --- INFO: Loading dataset.
+# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
+# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
+# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
+# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
+# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
+# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
+# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
+# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
+# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
+# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
+# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
+# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
+# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
+# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
+# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
+# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
+# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
+# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
+# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
+# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
+# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
+# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
+# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
+# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
+# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
+# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
+# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
+# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
+# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
+# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
+# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
+# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
+# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
+# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
+# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
+# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
+# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
+# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
+# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
+# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
+# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
+# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
+# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
+# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
+# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
+# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
+# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
+# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
+# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
+# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
+# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
+# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
+# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
+# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
+# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
+# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
+# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
+# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
+# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
+# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
+# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
+# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
+# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
+# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
+# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
+# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
+# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
+# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
+# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
+# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
+# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
+# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
+# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
+# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
+# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
+# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
+# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
+# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
+# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
+# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
+# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
+# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
+# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
+# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
+# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
+# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
+# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
+# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
+# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
+# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
+# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
+# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
+# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
+# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
+# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
+# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
+# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
+# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
+# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
+# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
+# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
+# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
+# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
+# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
+# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
+# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
+# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
+# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
+# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
+# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
+# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
+# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
+# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
+# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
+# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
+# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
+# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
+# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
+# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
+# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
+# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
+# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
+# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
+# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
+# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
+# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
+# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
+# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
+# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
+# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
+# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
+# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
+# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
+# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
+# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
+# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
+# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
+# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
+# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
+# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
+# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
+# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
+# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
+# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
+# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
+# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
+# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
+# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
+# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
+# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
+# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
+# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
+# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
+# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
+# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
+# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
+# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
+# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
+# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
+# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
+# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
+# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
+# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
+# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
+# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
+# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
+# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
+# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
+# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
+# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
+# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
+# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
+# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
+# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
+# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
+# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
+# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
+# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
+# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
+# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
+# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
+# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
+# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
+# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
+# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
+# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
+# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
+# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
+# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
+# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
+# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
+# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
+# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
+# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
+# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
+# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
+# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
+# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
+# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
+# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
+# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
+# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
+# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
+# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
+# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
+# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
+# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
+# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
+# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
+# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
+# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
+# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
+# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
+# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
+# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
+# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
+# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
+# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
+# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
+# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
+# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
+# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
+# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
+# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
+# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
+# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
+# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
+# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
+# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
+# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
+# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
+# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
+# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
+# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
+# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
+# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
+# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
+# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
+# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
+# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
+# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
+# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
+# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
+# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
+# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
+# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
+# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
+# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
+# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
+# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
+# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
+# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
+# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
+# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
+# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
+# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
+# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
+# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
+# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
+# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
+# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
+# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
+# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
+# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
+# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
+# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
+# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
+# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
+# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
+# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
+# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
+# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
+# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
+# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
+# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
+# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
+# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
+# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
+# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
+# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
+# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
+# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
+# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
+# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
+# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
+# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
+# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
+# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
+# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
+# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
+# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
+# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
+# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
+# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
+# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
+# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
+# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
+# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
+# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
+# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
+# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
+# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
+# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
+# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
+# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
+# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
+# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
+# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
+# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
+# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
+# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
+# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
+# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
+# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
+# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
+# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
+# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
+# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
+# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
+# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
+# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
+# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
+# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
+# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
+# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
+# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
+# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
+# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
+# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
+# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
+# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
+# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
+# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
+# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
+# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
+# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
+# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
+# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
+# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
+# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
+# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
+# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
+# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
+# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
+# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
+# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
+# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
+# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
+# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
+# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
+# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
+# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
+# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
+# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
+# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
+# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
+# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
+# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
+# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
+# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
+# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
+# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
+# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
+# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
+# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
+# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
+# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
+# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
+# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
+# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
+# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
+# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
+# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
+# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
+# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
+# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
+# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
+# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
+# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
+# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
+# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
+# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
+# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
+# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
+# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
+# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
+# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
+# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
+# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
+# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
+# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
+# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
+# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
+# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
+# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
+# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
+# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
+# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
+# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
+# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
+# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
+# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
+# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
+# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
+# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
+# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
+# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
+# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
+# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
+# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
+# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
+# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
+# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
+# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
+# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
+# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
+# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
+# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
+# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
+# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
+# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
+# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
+# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
+# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
+# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
+# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
+# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
+# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
+# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
+# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
+# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
+# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
+# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
+# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
+# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
+# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
+# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
+# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
+# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
+# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
+# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
+# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
+# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
+# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
+# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
+# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
+# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
+# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
+# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
+# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
+# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
+# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
+# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
+# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
+# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
+# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
+# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
+# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
+# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
+# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
+# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
+# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
+# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
+# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
+# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
+# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
+# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
+# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
+# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
+# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
+# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
+# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
+# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
+# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
+# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
+# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
+# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
+# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
+# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
+# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
+# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
+# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
+# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
+# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
+# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
+# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
+# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
+# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
+# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
+# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
+# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
+# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
+# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
+# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
+# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
+# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
+# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
+# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
+# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
+# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
+# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
+# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
+# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
+# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
+# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
+# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
+# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
+# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
+# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
+# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
+# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
+# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
+# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
+# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
+# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
+# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
+# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
+# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
+# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
+# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
+# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
+# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
+# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
+# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
+# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
+# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
+# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
+# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
+# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
+# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
+# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
+# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
+# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
+# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
+# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
+# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
+# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
+# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
+# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
+# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
+# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
+# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
+# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
+# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
+# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
+# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
+# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
+# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
+# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
+# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
+# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
+# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
+# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
+# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
+# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
+# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
+# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
+# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
+# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
+# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
+# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
+# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
+# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
+# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
+# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
+# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
+# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
+# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
+# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
+# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
+# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
+# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
+# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
+# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
+# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
+# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
+# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
+# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
+# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
+# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
+# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
+# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
+# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
+# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
+# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
+# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
+# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
+# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
+# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
+# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
+# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
+# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
+# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
+# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
+# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
+# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
+# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
+# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
+# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
+# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
+# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
+# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
+# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
+# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
+# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
+# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
+# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
+# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
+# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
+# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
+# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
+# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
+# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
+# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
+# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
+# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
+# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
+# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
+# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
+# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
+# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
+# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
+# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
+# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
+# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
+# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
+# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
+# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
+# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
+# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
+# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
+# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
+# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
+# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
+# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
+# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
+# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
+# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
+# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
+# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
+# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
+# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
+# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
+# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
+# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
+# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
+# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
+# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
+# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
+# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
+# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
+# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
+# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
+# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
+# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
+# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
+# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
+# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
+# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
+# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
+# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
+# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
+# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
+# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
+# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
+# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
+# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
+# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
+# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
+# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
+# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
+# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
+# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
+# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
+# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
+# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
+# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
+# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
+# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
+# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
+# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
+# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
+# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
+# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
+# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
+# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
+# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
+# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
+# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
+# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
+# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
+# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
+# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
+# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
+# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
+# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
+# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
+# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
+# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
+# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
+# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
+# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
+# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
+# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
+# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
+# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
+# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
+# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
+# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
+# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
+# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
+# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
+# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
+# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
+# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
+# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
+# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
+# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
+# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
+# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
+# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
+# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
+# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
+# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
+# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
+# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
+# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
+# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
+# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
+# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
+# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
+# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
+# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
+# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
+# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
+# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
+# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
+# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
+# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
+# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
+# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
+# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
+# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
+# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
+# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
+# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
+# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
+# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
+# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
+# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
+# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
+# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
+# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
+# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
+# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
+# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
+# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
+# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
+# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
+# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
+# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
+# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
+# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
+# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
+# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
+# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
+# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
+# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
+# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
+# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
+# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
+# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
+# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
+# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
+# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
+# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
+# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
+# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
+# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
+# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
+# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
+# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
+# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
+# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
+# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
+# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
+# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
+# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
+# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
+# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
+# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
+# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
+# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
+# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
+# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
+# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
+# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
+# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
+# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
+# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
+# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
+# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
+# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
+# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
+# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
+# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
+# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
+# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
+# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
+# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
+# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
+# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
+# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
+# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
+# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
+# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
+# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
+# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
+# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
+# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
+# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
+# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
+# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
+# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
+# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
+# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
+# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
+# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
+# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
+# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
+# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
+# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
+# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
+# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
+# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
+# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
+# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
+# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
+# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
+# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
+# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
+# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
+# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
+# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
+# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
+# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
+# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
+# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
+# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
+# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
+# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
+# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
+# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
+# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
+# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
+# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
+# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
+# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
+# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
+# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
+# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
+# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
+# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
+# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
+# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
+# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
+# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
+# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
+# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
+# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
+# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
+# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
+# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
+# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
+# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
+# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
+# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
+# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
+# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
+# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
+# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
+# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
+# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
+# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
+# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
+# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
+# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
+# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
+# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
+# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
+# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
+# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
+# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
+# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
+# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
+# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
+# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
+# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
+# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
+# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
+# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
+# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
+# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
+# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
+# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
+# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
+# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
+# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
+# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
+# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
+# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
+# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
+# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
+# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
+# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
+# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
+# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
+# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
+# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
+# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
+# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
+# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
+# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
+# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
+# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
+# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
+# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
+# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
+# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
+# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
+# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
+# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
+# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
+# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
+# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
+# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
+# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
+# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
+# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
+# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
+# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
+# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
+# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
+# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
+# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
+# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
+# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
+# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
+# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
+# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
+# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
+# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
+# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
+# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
+# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
+# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
+# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
+# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
+# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
+# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
+# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
+# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
+# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
+# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
+# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
+# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
+# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
+# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
+# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
+# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
+# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
+# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
+# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
+# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
+# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
+# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
+# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
+# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
+# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
+# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
+# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
+# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
+# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
+# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
+# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
+# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
+# 11:38:37 --- INFO: Training process end.
+# 11:38:37 --- INFO: Evaluating process start.
+# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
+# 11:38:37 --- INFO: Evaluating process end.
+# 11:38:37 --- INFO: Predicting process start.
+# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
+# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
+# 11:38:38 --- INFO: Predicting process end.
+# 11:38:38 --- INFO: --------------------- GNN end ----------------------
+# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
+# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
+# 11:38:38 --- INFO: Loading dataset.
+# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:38:39 --- INFO: Calculating NTScore for each niche.
+# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
+# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
+# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
+# 11:38:39 --- INFO: Output NTScore tables.
+# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
16.1.3 visualization
16.1.3.1 load ONTraC results
-
+
The NTScore and binarized niche cluster info were stored in cell metadata
-
-# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
-# <char> <char> <char> <char> <char> <char> <char> <int> <num>
-# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
-# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
-# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
-# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
-# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
-# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
+
+# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
+# <char> <char> <char> <char> <char> <char> <char> <int> <num>
+# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
+# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
+# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
+# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
+# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
+# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
The probability matrix of each cell assigned to each niche cluster and
connectivity between niche cluster were stored here.
-
-
+
+
16.1.3.2 niche cluster prob
-spatFeatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'niche cluster',
- expression_values = "prob",
- group_by = 'list_ID',
- feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
- point_border_col = 'gray'
- )
+spatFeatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'niche cluster',
+ expression_values = "prob",
+ group_by = 'list_ID',
+ feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
+ point_border_col = 'gray'
+ )
16.1.3.3 binarized niche cluster
-spatPlot2D(giotto_obj,
- spat_unit = "cell",
- group_by = 'list_ID',
- cell_color = "NicheCluster",
- color_as_factor = T,
- point_size = 1,
- point_border_stroke = NA,
- title="Niche cluster")
+spatPlot2D(giotto_obj,
+ spat_unit = "cell",
+ group_by = 'list_ID',
+ cell_color = "NicheCluster",
+ color_as_factor = T,
+ point_size = 1,
+ point_border_stroke = NA,
+ title="Niche cluster")
16.1.3.5 NT score
-spatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'rna',
- group_by = 'list_ID',
- cell_color = "NTScore",
- color_as_factor = FALSE,
- cell_color_gradient = "turbo",
- point_size = 1,
- point_border_stroke = NA
- )
+spatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ group_by = 'list_ID',
+ cell_color = "NTScore",
+ color_as_factor = FALSE,
+ cell_color_gradient = "turbo",
+ point_size = 1,
+ point_border_stroke = NA
+ )
-
+
diff --git a/interoperability-with-other-frameworks.html b/interoperability-with-other-frameworks.html
index 7050533..347d3af 100644
--- a/interoperability-with-other-frameworks.html
+++ b/interoperability-with-other-frameworks.html
@@ -521,10 +521,10 @@ 15 Interoperability with other fr
15.1 Load Giotto object
To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process:
-
+
Load a Giotto mini Visium object, which will be used for demonstrating interoperability.
-
+
15.2 Seurat
@@ -533,99 +533,99 @@ 15.2 Seurat
15.2.1 Conversion of Giotto Object to Seurat Object
To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function
-
+
Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data.
-gToS <- giottoToSeuratV5(gobject = gobject,
- spat_unit = "cell")
-
-plot1 <- VlnPlot(gToS,
- features = "nCount_rna",
- pt.size = 0.1) +
- NoLegend()
-
-plot2 <- SpatialFeaturePlot(gToS,
- features = "nCount_rna",
- pt.size.factor = 2) +
- theme(legend.position = "right")
-
-wrap_plots(plot1, plot2)
+gToS <- giottoToSeuratV5(gobject = gobject,
+ spat_unit = "cell")
+
+plot1 <- VlnPlot(gToS,
+ features = "nCount_rna",
+ pt.size = 0.1) +
+ NoLegend()
+
+plot2 <- SpatialFeaturePlot(gToS,
+ features = "nCount_rna",
+ pt.size.factor = 2) +
+ theme(legend.position = "right")
+
+wrap_plots(plot1, plot2)
15.2.1.1 Apply SCTransform
We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function.
-
+
15.2.1.2 Dimension Reduction
We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object:
-
+
15.2.2 Conversion of Seurat object Back to Giotto Object
To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks.
-giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
- spatial_assay = "rna",
- dim_reduction = c("pca", "umap"),
- sp_network = "Delaunay_network",
- nn_network = c("sNN.pca", "custom_NN" ))
+giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
+ spatial_assay = "rna",
+ dim_reduction = c("pca", "umap"),
+ sp_network = "Delaunay_network",
+ nn_network = c("sNN.pca", "custom_NN" ))
15.2.2.1 Clustering and Plotting UMAP
Here we perform K-means clustering on the UMAP results obtained from the Seurat object:
-## k-means clustering
-giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
- dim_reduction_to_use = "pca")
-
-#Plot UMAP post-clustering to visualize kmeans
-graph2 <- Giotto::plotUMAP(
- gobject = giottoFromSeurat,
- cell_color = "kmeans",
- show_NN_network = TRUE,
- point_size = 2.5
-)
+## k-means clustering
+giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
+ dim_reduction_to_use = "pca")
+
+#Plot UMAP post-clustering to visualize kmeans
+graph2 <- Giotto::plotUMAP(
+ gobject = giottoFromSeurat,
+ cell_color = "kmeans",
+ show_NN_network = TRUE,
+ point_size = 2.5
+)
15.2.2.2 Spatial CoExpression
We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function:
-ranktest <- binSpect(giottoFromSeurat,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "Delaunay_network")
-
-ext_spatial_genes <- ranktest[1:300,]$feats
-
-spat_cor_netw_DT <- detectSpatialCorFeats(
- giottoFromSeurat,
- method = "network",
- spatial_network_name = "Delaunay_network",
- subset_feats = ext_spatial_genes)
-
-top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
- feats = "Dsp",
- show_top_feats = 10)
-
-spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
- name = "spat_netw_clus",
- k = 7)
-heatmSpatialCorFeats(
- giottoFromSeurat,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL),
- save_plot = TRUE,
- show_plot = TRUE,
- return_plot = FALSE,
- save_param = list(base_height = 6, base_width = 8, units = 'cm'))
+ranktest <- binSpect(giottoFromSeurat,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "Delaunay_network")
+
+ext_spatial_genes <- ranktest[1:300,]$feats
+
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ giottoFromSeurat,
+ method = "network",
+ spatial_network_name = "Delaunay_network",
+ subset_feats = ext_spatial_genes)
+
+top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
+ feats = "Dsp",
+ show_top_feats = 10)
+
+spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
+ name = "spat_netw_clus",
+ k = 7)
+heatmSpatialCorFeats(
+ giottoFromSeurat,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ return_plot = FALSE,
+ save_param = list(base_height = 6, base_width = 8, units = 'cm'))
@@ -635,60 +635,60 @@ 15.3 SpatialExperiment
To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries.
-library(SpatialExperiment)
-library(ggspavis)
-library(pheatmap)
-library(scater)
-library(scran)
-library(nnSVG)
+library(SpatialExperiment)
+library(ggspavis)
+library(pheatmap)
+library(scater)
+library(scran)
+library(nnSVG)
15.3.1 Convert Giotto Object to SpatialExperiment Object
To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function.
-
+
The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use:
-
+
15.3.1.1 Identify top spatially variable genes with nnSVG
We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate:
-# One of the assays should be "logcounts"
-# We rename the normalized assay to "logcounts"
-assayNames(spe)[[2]] <- "logcounts"
-
-# Create model matrix for leiden clustering labels
-X <- model.matrix(~ colData(spe)$leiden_clus)
-dim(X)
+# One of the assays should be "logcounts"
+# We rename the normalized assay to "logcounts"
+assayNames(spe)[[2]] <- "logcounts"
+
+# Create model matrix for leiden clustering labels
+X <- model.matrix(~ colData(spe)$leiden_clus)
+dim(X)
- Run nnSVG
This step will take several minutes to run
-
+
15.3.2 Conversion of SpatialExperiment object back to Giotto
We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object.
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-print(giottoFromSPE)
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+print(giottoFromSPE)
15.3.2.1 Plotting top genes from nnSVG in Giotto
Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object.
-ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
- output = "data.table")
-
-ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
-spatFeatPlot2D(giottoFromSPE,
- expression_values = "scaled_rna_cell",
- feats = ext_spatial_genes[1:4],
- point_size = 2)
+ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
+ output = "data.table")
+
+ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
+spatFeatPlot2D(giottoFromSPE,
+ expression_values = "scaled_rna_cell",
+ feats = ext_spatial_genes[1:4],
+ point_size = 2)
@@ -701,97 +701,97 @@ 15.4 AnnData
15.4.1 Load Required Libraries
To start, we need to load the necessary libraries, including reticulate for interfacing with Python.
-
+
15.4.2 Specify Path for Results
First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions.
-# Specify path to which results may be saved
-results_directory <- "results/03_session4/giotto_anndata_conversion/"
-
-instrs <- showGiottoInstructions(gobject)
-
-mini_gobject <- replaceGiottoInstructions(gobject = gobject,
- instructions = instrs)
+# Specify path to which results may be saved
+results_directory <- "results/03_session4/giotto_anndata_conversion/"
+
+instrs <- showGiottoInstructions(gobject)
+
+mini_gobject <- replaceGiottoInstructions(gobject = gobject,
+ instructions = instrs)
15.4.2.1 Create Default kNN Network
We will create a k-nearest neighbor (kNN) network using mostly default parameters.
-
+
15.4.3 Giotto To AnnData
To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData()
-
+
Next, we import scanpy and perform a series of preprocessing steps on the AnnData object.
-scanpy <- import("scanpy")
-
-adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
-
-# Normalize total counts per cell
-scanpy$pp$normalize_total(adata, target_sum=1e4)
-
-# Log-transform the data
-scanpy$pp$log1p(adata)
-
-# Perform PCA
-scanpy$pp$pca(adata, n_comps=40L)
-
-# Compute the neighborhood graph
-scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
-
-# Run UMAP
-scanpy$tl$umap(adata)
-
-# Save the processed AnnData object
-adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
-
-processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
+scanpy <- import("scanpy")
+
+adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
+
+# Normalize total counts per cell
+scanpy$pp$normalize_total(adata, target_sum=1e4)
+
+# Log-transform the data
+scanpy$pp$log1p(adata)
+
+# Perform PCA
+scanpy$pp$pca(adata, n_comps=40L)
+
+# Compute the neighborhood graph
+scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
+
+# Run UMAP
+scanpy$tl$umap(adata)
+
+# Save the processed AnnData object
+adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
+
+processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
15.4.4 Convert AnnData to Giotto
Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto.
-
+
15.4.4.1 UMAP Visualization
Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object.
-Giotto::plotUMAP(
- gobject = giottoFromAnndata,
- dim_reduction_name = "umap.ad",
- cell_color = "leiden_clus",
- point_size = 3
-)
+Giotto::plotUMAP(
+ gobject = giottoFromAnndata,
+ dim_reduction_name = "umap.ad",
+ cell_color = "leiden_clus",
+ point_size = 3
+)
15.5 Create mini Vizgen object
-mini_gobject <- loadGiottoMini(dataset = "vizgen",
- python_path = NULL)
-
-mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
- instructions = instrs)
-mini_gobject <- createNearestNetwork(gobject = mini_gobject,
- spat_unit = "aggregate",
- feat_type = "rna",
- type = "kNN",
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap",
- k = 6,
- name = "new_network")
+mini_gobject <- loadGiottoMini(dataset = "vizgen",
+ python_path = NULL)
+
+mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
+ instructions = instrs)
+mini_gobject <- createNearestNetwork(gobject = mini_gobject,
+ spat_unit = "aggregate",
+ feat_type = "rna",
+ type = "kNN",
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap",
+ k = 6,
+ name = "new_network")
Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details.
-
+
diff --git a/introduction-to-the-giotto-package.html b/introduction-to-the-giotto-package.html
index e22d6d1..0ab2014 100644
--- a/introduction-to-the-giotto-package.html
+++ b/introduction-to-the-giotto-package.html
@@ -546,19 +546,58 @@ 4.3 Installation + python environ
4.3.1 Giotto installation
Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository.
Much of this already covered in Section 2.2, but the highlights are:
-
-- Ensure your system
-
-To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak.
+
+4.3.1.1 System prerequisites
+
+- for windows, Rtools needs to be installed
+- a major dependency terra needs
GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux
+
+
+
+4.3.1.2 Installation of released version
+To install the currently released version of Giotto in a single step:
+This should automatically install all the Giotto dependencies and other Giotto module packages.
+
+
+4.3.1.3 Installation of dev branch Giotto packages
+You can install dev branch versions by using devtools::install_github()
+Core module dev branchs:
+
+- “drieslab/Giotto@suite_dev”
+- “drieslab/GiottoVisuals@dev”
+- “drieslab/GiottoClass@dev”
+- “drieslab/GiottoUtils@dev”
+
+
+pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages.
+
+
+4.3.1.4 Common install issues
+If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work.
+
+If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it.
+
+## # Downloading packages -------------------------------------------------------
+## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s]
+## Successfully downloaded 1 package in 1.2 seconds.
+##
+## The following package(s) will be installed:
+## - irlba [2.3.5.1]
+## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu".
+##
+## # Installing packages --------------------------------------------------------
+## - Installing irlba ... OK [built from source and cached in 5.7s]
+## Successfully installed 1 package in 5.7 seconds.
+
spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.5,
- point_shape = 'no_border',
- background_color = 'black',
- save_plot = F,
- show_legend = TRUE)
-
-spatPlot2D(gobject = subcell_v_brain,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat'))spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.5,
+ point_shape = 'no_border',
+ background_color = 'black',
+ save_plot = F,
+ show_legend = TRUE)
+
+spatPlot2D(gobject = subcell_v_brain,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat'))Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) @@ -918,37 +918,37 @@
17.7.5 Visualizing clustering
17.7.6 Cropping objects
We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below.
-subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
- spat_unit = ":all:",
- x_min = 4000,
- x_max = 7000,
- y_min = -6500,
- y_max = -3500,
- z_max = NULL,
- z_min = NULL)
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 2,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_vis_spat_crop'))
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat_crop'))
-
+subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
+ spat_unit = ":all:",
+ x_min = 4000,
+ x_max = 7000,
+ y_min = -6500,
+ y_max = -3500,
+ z_max = NULL,
+ z_min = NULL)
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 2,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_vis_spat_crop'))
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat_crop'))
+
Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right)
diff --git a/interoperability-with-isolated-tools.html b/interoperability-with-isolated-tools.html
index ceb2d0a..8cca7e6 100644
--- a/interoperability-with-isolated-tools.html
+++ b/interoperability-with-isolated-tools.html
@@ -526,1521 +526,1521 @@
16.1.1 Dataset downloaddata_path <- "data"
-dir.create(data_path)
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
- destfile = file.path(data_path,"counts.h5ad"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
- destfile = file.path(data_path,"cell_labels.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
+data_path <- "data"
+dir.create(data_path)
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
+ destfile = file.path(data_path,"counts.h5ad"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
+ destfile = file.path(data_path,"cell_labels.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
16.1.2 Create the Giotto object
-library(Giotto)
-library(reticulate)
-
-## Set instructions
-results_folder <- "results"
-dir.create(results_folder)
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-
-## load data
-
-### meta data
-meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-colnames(meta_df)[[1]] <- "cell_ID"
-slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
-slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
-
-### counts matrix
-scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
-ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
-reticulate::source_python(ad2g_path)
-adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
-X <- extract_expression(adata)
-cID <- extract_cell_IDs(adata)
-fID <- extract_feat_IDs(adata)
-rownames(X) <- fID
-colnames(X) <- cID
-slice1_filter <- cID %in% slice1_meta_df$cell_ID
-slice2_filter <- cID %in% slice2_meta_df$cell_ID
-slice1_exp <- X[,slice1_filter]
-slice2_exp <- X[,slice2_filter]
-
-### cell segmentation to cell location
-segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_df <- rbind(segments_1_df, segments_2_df)
-loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
-loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
-loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
-centr.x <- apply(loc.use[,loc.x],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-centr.y <- apply(loc.use[,loc.y],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
-slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
-slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
-
-## giotto obj
-giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
- spatial_locs = slice1_spatial_locs_df,
- cell_metadata = slice1_meta_df)
-giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
- spatial_locs = slice2_spatial_locs_df,
- cell_metadata = slice2_meta_df)
-giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
- gobject_names = c('mouse2_slice229',
- 'mouse2_slice300'), # name for each samples
- join_method = 'z_stack')
-# We skip the processing process here to save time and use the given cell type
-# annotation directly
-
-ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
- cell_type = "subclass",
- output_path = data_path,
- spat_unit = "cell",
- feat_type = "rna",
- verbose = TRUE)
-
-# Cell_ID Sample x y Cell_Type
-# <char> <char> <dbl> <dbl> <char>
-# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
-# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
-# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
-# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
-# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
-# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
+library(Giotto)
+library(reticulate)
+
+## Set instructions
+results_folder <- "results"
+dir.create(results_folder)
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+
+## load data
+
+### meta data
+meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+colnames(meta_df)[[1]] <- "cell_ID"
+slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
+slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
+
+### counts matrix
+scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
+ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
+reticulate::source_python(ad2g_path)
+adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
+X <- extract_expression(adata)
+cID <- extract_cell_IDs(adata)
+fID <- extract_feat_IDs(adata)
+rownames(X) <- fID
+colnames(X) <- cID
+slice1_filter <- cID %in% slice1_meta_df$cell_ID
+slice2_filter <- cID %in% slice2_meta_df$cell_ID
+slice1_exp <- X[,slice1_filter]
+slice2_exp <- X[,slice2_filter]
+
+### cell segmentation to cell location
+segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_df <- rbind(segments_1_df, segments_2_df)
+loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
+loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
+loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
+centr.x <- apply(loc.use[,loc.x],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+centr.y <- apply(loc.use[,loc.y],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
+slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
+slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
+
+## giotto obj
+giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
+ spatial_locs = slice1_spatial_locs_df,
+ cell_metadata = slice1_meta_df)
+giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
+ spatial_locs = slice2_spatial_locs_df,
+ cell_metadata = slice2_meta_df)
+giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
+ gobject_names = c('mouse2_slice229',
+ 'mouse2_slice300'), # name for each samples
+ join_method = 'z_stack')
+# We skip the processing process here to save time and use the given cell type
+# annotation directly
+
+ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
+ cell_type = "subclass",
+ output_path = data_path,
+ spat_unit = "cell",
+ feat_type = "rna",
+ verbose = TRUE)
+
+# Cell_ID Sample x y Cell_Type
+# <char> <char> <dbl> <dbl> <char>
+# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
+# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
+# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
+# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
+# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
+# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro
Spatial cell type distributions
-spatPlot2D(giotto_obj,
- group_by = "slice_id",
- cell_color = "subclass",
- point_size = 1,
- point_border_stroke = NA,
- legend_text = 6)
+spatPlot2D(giotto_obj,
+ group_by = "slice_id",
+ cell_color = "subclass",
+ point_size = 1,
+ point_border_stroke = NA,
+ legend_text = 6)
16.1.2.1 Installation ONTraC
You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node.
It will run for less than 10 minutes on this example dataset.
For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time.
-source ~/.bash_profile
-conda create -y -n ONTraC python=3.11
-conda activate ONTraC
-pip install git+https://github.com/gyuanlab/ONTraC.git@V2
-# Retrieving notices: ...working... done
-# Channels:
-# - conda-forge
-# - bioconda
-# - r
-# - pytorch
-# - defaults
-# Platform: osx-arm64
-# Collecting package metadata (repodata.json): ...working... done
-# Solving environment: ...working... done
-#
-# ## Package Plan ##
-#
-# environment location: /Users/wwang/miniconda3/envs/ONTraC
-#
-# added / updated specs:
-# - python=3.11
-#
-#
-# The following packages will be downloaded:
-#
-# package | build
-# ---------------------------|-----------------
-# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
-# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
-# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
-# ------------------------------------------------------------
-# Total: 4.3 MB
-#
-# The following NEW packages will be INSTALLED:
-#
-# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
-# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
-# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
-# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
-# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
-# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
-# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
-# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
-# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
-# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
-# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
-# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
-# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
-# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
-# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
-# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
-#
-#
-#
-# Downloading and Extracting Packages: ...working... done
-# Preparing transaction: ...working... done
-# Verifying transaction: ...working... done
-# Executing transaction: ...working... done
-# #
-# # To activate this environment, use
-# #
-# # $ conda activate ONTraC
-# #
-# # To deactivate an active environment, use
-# #
-# # $ conda deactivate
-#
-# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
-# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
-# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
-# Running command git checkout -b V2 --track origin/V2
-# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
-# Installing build dependencies: started
-# Switched to a new branch 'V2'
-# branch 'V2' set up to track 'origin/V2'.
-# Installing build dependencies: finished with status 'done'
-# Getting requirements to build wheel: started
-# Getting requirements to build wheel: finished with status 'done'
-# Preparing metadata (pyproject.toml): started
-# Preparing metadata (pyproject.toml): finished with status 'done'
-# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
-# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
-# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
-# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
-# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
-# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
-# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
-# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
-# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
-# Collecting session-info (from ONTraC==2.0a9.dev7)
-# Using cached session_info-1.0.0-py3-none-any.whl
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
-# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
-# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
-# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
-# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
-# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
-# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
-# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
-# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
-# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
-# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
-# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
-# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
-# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
-# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
-# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
-# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
-# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
-# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
-# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
-# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
-# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
-# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
-# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
-# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
-# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
-# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
-# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
-# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
-# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
-# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
-# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
-# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
-# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
-# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
-# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
-# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
-# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
-# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
-# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
-# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
-# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
-# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
-# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
-# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
-# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
-# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
-# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
-# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
-# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
-# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
-# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
-# Using cached idna-3.7-py3-none-any.whl (66 kB)
-# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
-# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
-# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
-# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
-# Building wheels for collected packages: ONTraC
-# Building wheel for ONTraC (pyproject.toml): started
-# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
-# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
-# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
-# Successfully built ONTraC
-# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
-# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
+source ~/.bash_profile
+conda create -y -n ONTraC python=3.11
+conda activate ONTraC
+pip install git+https://github.com/gyuanlab/ONTraC.git@V2
+# Retrieving notices: ...working... done
+# Channels:
+# - conda-forge
+# - bioconda
+# - r
+# - pytorch
+# - defaults
+# Platform: osx-arm64
+# Collecting package metadata (repodata.json): ...working... done
+# Solving environment: ...working... done
+#
+# ## Package Plan ##
+#
+# environment location: /Users/wwang/miniconda3/envs/ONTraC
+#
+# added / updated specs:
+# - python=3.11
+#
+#
+# The following packages will be downloaded:
+#
+# package | build
+# ---------------------------|-----------------
+# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
+# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
+# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
+# ------------------------------------------------------------
+# Total: 4.3 MB
+#
+# The following NEW packages will be INSTALLED:
+#
+# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
+# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
+# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
+# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
+# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
+# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
+# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
+# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
+# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
+# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
+# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
+# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
+# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
+# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
+# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
+# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
+#
+#
+#
+# Downloading and Extracting Packages: ...working... done
+# Preparing transaction: ...working... done
+# Verifying transaction: ...working... done
+# Executing transaction: ...working... done
+# #
+# # To activate this environment, use
+# #
+# # $ conda activate ONTraC
+# #
+# # To deactivate an active environment, use
+# #
+# # $ conda deactivate
+#
+# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
+# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
+# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
+# Running command git checkout -b V2 --track origin/V2
+# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
+# Installing build dependencies: started
+# Switched to a new branch 'V2'
+# branch 'V2' set up to track 'origin/V2'.
+# Installing build dependencies: finished with status 'done'
+# Getting requirements to build wheel: started
+# Getting requirements to build wheel: finished with status 'done'
+# Preparing metadata (pyproject.toml): started
+# Preparing metadata (pyproject.toml): finished with status 'done'
+# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
+# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
+# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
+# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
+# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
+# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
+# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
+# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
+# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
+# Collecting session-info (from ONTraC==2.0a9.dev7)
+# Using cached session_info-1.0.0-py3-none-any.whl
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
+# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
+# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
+# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
+# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
+# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
+# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
+# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
+# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
+# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
+# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
+# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
+# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
+# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
+# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
+# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
+# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
+# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
+# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
+# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
+# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
+# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
+# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
+# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
+# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
+# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
+# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
+# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
+# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
+# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
+# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
+# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
+# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
+# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
+# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
+# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
+# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
+# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
+# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
+# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
+# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
+# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
+# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
+# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
+# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
+# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
+# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
+# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
+# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
+# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
+# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
+# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
+# Using cached idna-3.7-py3-none-any.whl (66 kB)
+# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
+# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
+# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
+# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
+# Building wheels for collected packages: ONTraC
+# Building wheel for ONTraC (pyproject.toml): started
+# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
+# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
+# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
+# Successfully built ONTraC
+# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
+# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4
16.1.2.2 Running ONTraC
-
-source ~/.bash_profile
-conda activate ONTraC_v2_py311
-ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
-# ##################################################################################
-#
-# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
-# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
-# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
-# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
-# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
-#
-# version: 2.0a11
-#
-# ##################################################################################
-# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
-# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
-# 11:33:23 --- INFO: -------- I/O options -------
-# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
-# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
-# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
-# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
-# 11:33:23 --- INFO: -------- preprocessing options -------
-# 11:33:23 --- INFO: resolution: 10.0
-# 11:33:23 --- INFO: -------- niche net constr options -------
-# 11:33:23 --- INFO: n_cpu: 4
-# 11:33:23 --- INFO: n_neighbors: 50
-# 11:33:23 --- INFO: n_local: 20
-# 11:33:23 --- INFO: embedding_adjust: False
-# 11:33:23 --- INFO: sigma: 1
-# 11:33:23 --- INFO: -------- train options -------
-# 11:33:23 --- INFO: device: cpu
-# 11:33:23 --- INFO: epochs: 1000
-# 11:33:23 --- INFO: batch_size: 0
-# 11:33:23 --- INFO: patience: 100
-# 11:33:23 --- INFO: min_delta: 0.001
-# 11:33:23 --- INFO: min_epochs: 50
-# 11:33:23 --- INFO: seed: 42
-# 11:33:23 --- INFO: lr: 0.03
-# 11:33:23 --- INFO: hidden_feats: 4
-# 11:33:23 --- INFO: k: 6
-# 11:33:23 --- INFO: modularity_loss_weight: 1.0
-# 11:33:23 --- INFO: purity_loss_weight: 100.0
-# 11:33:23 --- INFO: regularization_loss_weight: 0.1
-# 11:33:23 --- INFO: beta: 0.3
-# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
-# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
-# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
-# 11:33:23 --- INFO: ------------- Niche network construct ---------------
-# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
-# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
-# 11:33:28 --- INFO: Generating samples.yaml file.
-# 11:33:28 --- INFO: ------------ Niche network construct end ------------
-# 11:33:28 --- INFO: ------------------------ GNN ------------------------
-# 11:33:28 --- INFO: Loading dataset.
-# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
-# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
-# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
-# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
-# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
-# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
-# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
-# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
-# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
-# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
-# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
-# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
-# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
-# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
-# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
-# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
-# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
-# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
-# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
-# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
-# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
-# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
-# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
-# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
-# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
-# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
-# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
-# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
-# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
-# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
-# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
-# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
-# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
-# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
-# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
-# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
-# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
-# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
-# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
-# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
-# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
-# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
-# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
-# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
-# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
-# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
-# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
-# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
-# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
-# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
-# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
-# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
-# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
-# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
-# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
-# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
-# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
-# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
-# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
-# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
-# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
-# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
-# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
-# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
-# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
-# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
-# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
-# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
-# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
-# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
-# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
-# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
-# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
-# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
-# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
-# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
-# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
-# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
-# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
-# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
-# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
-# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
-# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
-# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
-# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
-# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
-# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
-# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
-# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
-# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
-# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
-# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
-# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
-# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
-# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
-# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
-# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
-# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
-# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
-# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
-# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
-# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
-# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
-# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
-# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
-# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
-# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
-# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
-# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
-# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
-# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
-# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
-# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
-# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
-# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
-# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
-# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
-# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
-# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
-# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
-# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
-# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
-# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
-# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
-# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
-# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
-# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
-# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
-# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
-# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
-# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
-# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
-# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
-# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
-# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
-# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
-# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
-# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
-# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
-# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
-# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
-# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
-# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
-# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
-# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
-# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
-# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
-# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
-# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
-# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
-# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
-# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
-# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
-# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
-# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
-# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
-# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
-# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
-# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
-# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
-# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
-# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
-# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
-# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
-# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
-# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
-# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
-# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
-# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
-# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
-# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
-# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
-# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
-# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
-# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
-# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
-# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
-# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
-# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
-# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
-# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
-# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
-# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
-# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
-# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
-# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
-# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
-# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
-# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
-# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
-# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
-# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
-# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
-# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
-# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
-# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
-# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
-# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
-# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
-# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
-# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
-# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
-# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
-# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
-# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
-# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
-# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
-# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
-# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
-# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
-# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
-# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
-# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
-# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
-# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
-# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
-# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
-# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
-# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
-# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
-# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
-# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
-# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
-# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
-# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
-# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
-# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
-# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
-# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
-# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
-# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
-# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
-# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
-# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
-# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
-# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
-# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
-# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
-# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
-# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
-# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
-# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
-# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
-# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
-# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
-# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
-# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
-# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
-# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
-# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
-# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
-# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
-# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
-# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
-# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
-# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
-# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
-# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
-# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
-# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
-# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
-# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
-# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
-# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
-# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
-# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
-# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
-# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
-# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
-# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
-# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
-# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
-# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
-# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
-# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
-# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
-# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
-# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
-# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
-# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
-# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
-# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
-# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
-# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
-# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
-# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
-# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
-# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
-# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
-# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
-# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
-# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
-# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
-# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
-# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
-# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
-# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
-# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
-# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
-# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
-# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
-# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
-# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
-# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
-# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
-# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
-# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
-# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
-# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
-# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
-# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
-# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
-# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
-# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
-# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
-# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
-# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
-# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
-# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
-# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
-# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
-# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
-# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
-# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
-# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
-# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
-# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
-# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
-# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
-# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
-# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
-# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
-# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
-# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
-# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
-# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
-# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
-# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
-# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
-# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
-# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
-# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
-# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
-# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
-# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
-# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
-# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
-# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
-# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
-# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
-# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
-# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
-# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
-# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
-# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
-# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
-# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
-# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
-# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
-# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
-# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
-# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
-# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
-# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
-# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
-# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
-# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
-# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
-# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
-# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
-# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
-# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
-# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
-# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
-# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
-# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
-# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
-# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
-# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
-# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
-# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
-# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
-# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
-# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
-# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
-# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
-# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
-# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
-# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
-# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
-# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
-# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
-# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
-# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
-# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
-# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
-# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
-# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
-# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
-# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
-# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
-# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
-# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
-# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
-# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
-# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
-# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
-# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
-# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
-# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
-# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
-# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
-# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
-# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
-# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
-# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
-# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
-# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
-# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
-# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
-# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
-# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
-# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
-# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
-# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
-# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
-# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
-# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
-# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
-# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
-# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
-# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
-# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
-# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
-# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
-# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
-# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
-# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
-# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
-# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
-# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
-# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
-# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
-# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
-# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
-# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
-# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
-# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
-# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
-# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
-# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
-# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
-# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
-# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
-# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
-# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
-# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
-# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
-# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
-# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
-# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
-# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
-# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
-# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
-# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
-# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
-# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
-# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
-# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
-# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
-# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
-# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
-# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
-# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
-# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
-# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
-# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
-# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
-# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
-# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
-# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
-# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
-# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
-# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
-# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
-# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
-# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
-# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
-# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
-# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
-# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
-# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
-# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
-# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
-# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
-# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
-# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
-# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
-# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
-# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
-# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
-# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
-# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
-# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
-# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
-# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
-# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
-# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
-# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
-# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
-# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
-# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
-# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
-# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
-# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
-# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
-# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
-# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
-# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
-# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
-# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
-# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
-# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
-# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
-# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
-# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
-# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
-# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
-# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
-# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
-# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
-# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
-# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
-# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
-# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
-# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
-# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
-# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
-# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
-# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
-# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
-# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
-# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
-# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
-# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
-# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
-# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
-# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
-# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
-# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
-# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
-# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
-# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
-# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
-# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
-# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
-# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
-# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
-# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
-# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
-# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
-# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
-# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
-# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
-# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
-# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
-# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
-# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
-# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
-# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
-# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
-# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
-# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
-# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
-# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
-# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
-# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
-# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
-# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
-# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
-# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
-# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
-# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
-# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
-# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
-# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
-# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
-# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
-# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
-# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
-# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
-# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
-# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
-# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
-# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
-# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
-# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
-# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
-# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
-# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
-# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
-# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
-# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
-# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
-# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
-# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
-# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
-# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
-# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
-# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
-# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
-# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
-# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
-# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
-# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
-# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
-# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
-# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
-# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
-# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
-# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
-# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
-# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
-# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
-# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
-# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
-# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
-# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
-# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
-# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
-# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
-# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
-# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
-# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
-# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
-# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
-# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
-# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
-# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
-# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
-# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
-# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
-# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
-# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
-# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
-# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
-# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
-# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
-# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
-# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
-# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
-# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
-# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
-# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
-# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
-# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
-# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
-# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
-# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
-# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
-# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
-# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
-# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
-# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
-# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
-# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
-# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
-# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
-# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
-# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
-# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
-# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
-# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
-# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
-# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
-# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
-# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
-# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
-# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
-# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
-# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
-# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
-# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
-# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
-# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
-# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
-# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
-# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
-# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
-# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
-# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
-# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
-# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
-# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
-# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
-# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
-# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
-# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
-# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
-# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
-# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
-# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
-# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
-# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
-# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
-# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
-# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
-# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
-# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
-# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
-# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
-# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
-# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
-# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
-# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
-# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
-# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
-# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
-# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
-# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
-# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
-# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
-# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
-# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
-# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
-# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
-# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
-# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
-# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
-# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
-# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
-# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
-# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
-# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
-# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
-# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
-# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
-# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
-# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
-# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
-# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
-# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
-# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
-# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
-# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
-# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
-# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
-# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
-# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
-# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
-# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
-# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
-# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
-# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
-# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
-# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
-# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
-# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
-# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
-# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
-# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
-# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
-# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
-# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
-# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
-# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
-# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
-# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
-# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
-# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
-# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
-# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
-# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
-# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
-# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
-# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
-# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
-# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
-# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
-# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
-# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
-# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
-# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
-# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
-# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
-# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
-# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
-# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
-# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
-# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
-# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
-# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
-# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
-# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
-# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
-# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
-# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
-# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
-# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
-# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
-# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
-# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
-# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
-# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
-# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
-# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
-# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
-# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
-# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
-# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
-# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
-# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
-# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
-# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
-# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
-# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
-# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
-# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
-# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
-# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
-# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
-# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
-# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
-# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
-# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
-# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
-# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
-# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
-# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
-# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
-# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
-# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
-# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
-# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
-# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
-# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
-# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
-# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
-# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
-# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
-# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
-# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
-# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
-# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
-# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
-# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
-# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
-# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
-# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
-# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
-# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
-# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
-# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
-# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
-# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
-# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
-# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
-# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
-# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
-# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
-# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
-# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
-# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
-# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
-# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
-# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
-# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
-# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
-# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
-# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
-# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
-# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
-# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
-# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
-# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
-# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
-# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
-# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
-# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
-# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
-# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
-# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
-# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
-# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
-# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
-# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
-# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
-# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
-# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
-# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
-# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
-# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
-# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
-# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
-# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
-# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
-# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
-# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
-# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
-# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
-# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
-# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
-# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
-# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
-# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
-# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
-# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
-# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
-# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
-# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
-# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
-# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
-# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
-# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
-# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
-# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
-# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
-# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
-# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
-# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
-# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
-# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
-# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
-# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
-# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
-# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
-# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
-# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
-# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
-# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
-# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
-# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
-# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
-# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
-# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
-# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
-# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
-# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
-# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
-# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
-# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
-# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
-# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
-# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
-# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
-# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
-# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
-# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
-# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
-# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
-# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
-# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
-# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
-# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
-# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
-# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
-# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
-# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
-# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
-# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
-# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
-# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
-# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
-# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
-# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
-# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
-# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
-# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
-# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
-# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
-# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
-# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
-# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
-# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
-# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
-# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
-# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
-# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
-# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
-# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
-# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
-# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
-# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
-# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
-# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
-# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
-# 11:38:37 --- INFO: Training process end.
-# 11:38:37 --- INFO: Evaluating process start.
-# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
-# 11:38:37 --- INFO: Evaluating process end.
-# 11:38:37 --- INFO: Predicting process start.
-# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
-# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
-# 11:38:38 --- INFO: Predicting process end.
-# 11:38:38 --- INFO: --------------------- GNN end ----------------------
-# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
-# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
-# 11:38:38 --- INFO: Loading dataset.
-# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:38:39 --- INFO: Calculating NTScore for each niche.
-# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
-# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
-# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
-# 11:38:39 --- INFO: Output NTScore tables.
-# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
+
+source ~/.bash_profile
+conda activate ONTraC_v2_py311
+ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log
+# ##################################################################################
+#
+# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
+# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
+# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
+# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
+# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
+#
+# version: 2.0a11
+#
+# ##################################################################################
+# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
+# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
+# 11:33:23 --- INFO: -------- I/O options -------
+# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
+# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
+# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
+# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
+# 11:33:23 --- INFO: -------- preprocessing options -------
+# 11:33:23 --- INFO: resolution: 10.0
+# 11:33:23 --- INFO: -------- niche net constr options -------
+# 11:33:23 --- INFO: n_cpu: 4
+# 11:33:23 --- INFO: n_neighbors: 50
+# 11:33:23 --- INFO: n_local: 20
+# 11:33:23 --- INFO: embedding_adjust: False
+# 11:33:23 --- INFO: sigma: 1
+# 11:33:23 --- INFO: -------- train options -------
+# 11:33:23 --- INFO: device: cpu
+# 11:33:23 --- INFO: epochs: 1000
+# 11:33:23 --- INFO: batch_size: 0
+# 11:33:23 --- INFO: patience: 100
+# 11:33:23 --- INFO: min_delta: 0.001
+# 11:33:23 --- INFO: min_epochs: 50
+# 11:33:23 --- INFO: seed: 42
+# 11:33:23 --- INFO: lr: 0.03
+# 11:33:23 --- INFO: hidden_feats: 4
+# 11:33:23 --- INFO: k: 6
+# 11:33:23 --- INFO: modularity_loss_weight: 1.0
+# 11:33:23 --- INFO: purity_loss_weight: 100.0
+# 11:33:23 --- INFO: regularization_loss_weight: 0.1
+# 11:33:23 --- INFO: beta: 0.3
+# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
+# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
+# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
+# 11:33:23 --- INFO: ------------- Niche network construct ---------------
+# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
+# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
+# 11:33:28 --- INFO: Generating samples.yaml file.
+# 11:33:28 --- INFO: ------------ Niche network construct end ------------
+# 11:33:28 --- INFO: ------------------------ GNN ------------------------
+# 11:33:28 --- INFO: Loading dataset.
+# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
+# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
+# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
+# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
+# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
+# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
+# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
+# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
+# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
+# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
+# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
+# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
+# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
+# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
+# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
+# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
+# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
+# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
+# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
+# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
+# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
+# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
+# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
+# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
+# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
+# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
+# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
+# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
+# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
+# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
+# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
+# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
+# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
+# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
+# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
+# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
+# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
+# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
+# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
+# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
+# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
+# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
+# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
+# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
+# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
+# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
+# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
+# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
+# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
+# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
+# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
+# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
+# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
+# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
+# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
+# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
+# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
+# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
+# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
+# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
+# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
+# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
+# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
+# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
+# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
+# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
+# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
+# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
+# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
+# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
+# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
+# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
+# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
+# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
+# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
+# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
+# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
+# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
+# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
+# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
+# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
+# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
+# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
+# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
+# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
+# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
+# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
+# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
+# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
+# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
+# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
+# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
+# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
+# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
+# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
+# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
+# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
+# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
+# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
+# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
+# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
+# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
+# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
+# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
+# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
+# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
+# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
+# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
+# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
+# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
+# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
+# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
+# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
+# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
+# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
+# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
+# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
+# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
+# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
+# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
+# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
+# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
+# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
+# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
+# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
+# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
+# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
+# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
+# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
+# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
+# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
+# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
+# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
+# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
+# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
+# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
+# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
+# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
+# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
+# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
+# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
+# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
+# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
+# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
+# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
+# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
+# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
+# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
+# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
+# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
+# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
+# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
+# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
+# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
+# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
+# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
+# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
+# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
+# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
+# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
+# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
+# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
+# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
+# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
+# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
+# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
+# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
+# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
+# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
+# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
+# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
+# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
+# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
+# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
+# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
+# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
+# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
+# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
+# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
+# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
+# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
+# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
+# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
+# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
+# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
+# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
+# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
+# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
+# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
+# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
+# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
+# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
+# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
+# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
+# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
+# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
+# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
+# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
+# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
+# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
+# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
+# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
+# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
+# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
+# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
+# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
+# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
+# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
+# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
+# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
+# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
+# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
+# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
+# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
+# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
+# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
+# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
+# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
+# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
+# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
+# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
+# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
+# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
+# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
+# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
+# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
+# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
+# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
+# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
+# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
+# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
+# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
+# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
+# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
+# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
+# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
+# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
+# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
+# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
+# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
+# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
+# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
+# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
+# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
+# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
+# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
+# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
+# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
+# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
+# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
+# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
+# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
+# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
+# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
+# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
+# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
+# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
+# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
+# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
+# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
+# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
+# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
+# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
+# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
+# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
+# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
+# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
+# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
+# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
+# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
+# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
+# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
+# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
+# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
+# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
+# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
+# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
+# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
+# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
+# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
+# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
+# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
+# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
+# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
+# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
+# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
+# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
+# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
+# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
+# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
+# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
+# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
+# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
+# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
+# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
+# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
+# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
+# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
+# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
+# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
+# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
+# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
+# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
+# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
+# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
+# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
+# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
+# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
+# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
+# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
+# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
+# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
+# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
+# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
+# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
+# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
+# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
+# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
+# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
+# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
+# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
+# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
+# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
+# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
+# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
+# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
+# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
+# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
+# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
+# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
+# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
+# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
+# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
+# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
+# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
+# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
+# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
+# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
+# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
+# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
+# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
+# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
+# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
+# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
+# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
+# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
+# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
+# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
+# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
+# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
+# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
+# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
+# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
+# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
+# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
+# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
+# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
+# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
+# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
+# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
+# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
+# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
+# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
+# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
+# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
+# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
+# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
+# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
+# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
+# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
+# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
+# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
+# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
+# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
+# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
+# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
+# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
+# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
+# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
+# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
+# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
+# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
+# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
+# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
+# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
+# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
+# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
+# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
+# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
+# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
+# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
+# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
+# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
+# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
+# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
+# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
+# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
+# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
+# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
+# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
+# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
+# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
+# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
+# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
+# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
+# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
+# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
+# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
+# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
+# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
+# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
+# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
+# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
+# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
+# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
+# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
+# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
+# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
+# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
+# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
+# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
+# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
+# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
+# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
+# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
+# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
+# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
+# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
+# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
+# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
+# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
+# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
+# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
+# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
+# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
+# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
+# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
+# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
+# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
+# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
+# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
+# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
+# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
+# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
+# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
+# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
+# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
+# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
+# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
+# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
+# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
+# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
+# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
+# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
+# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
+# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
+# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
+# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
+# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
+# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
+# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
+# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
+# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
+# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
+# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
+# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
+# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
+# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
+# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
+# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
+# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
+# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
+# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
+# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
+# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
+# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
+# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
+# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
+# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
+# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
+# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
+# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
+# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
+# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
+# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
+# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
+# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
+# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
+# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
+# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
+# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
+# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
+# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
+# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
+# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
+# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
+# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
+# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
+# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
+# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
+# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
+# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
+# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
+# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
+# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
+# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
+# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
+# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
+# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
+# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
+# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
+# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
+# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
+# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
+# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
+# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
+# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
+# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
+# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
+# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
+# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
+# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
+# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
+# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
+# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
+# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
+# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
+# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
+# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
+# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
+# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
+# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
+# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
+# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
+# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
+# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
+# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
+# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
+# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
+# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
+# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
+# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
+# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
+# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
+# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
+# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
+# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
+# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
+# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
+# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
+# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
+# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
+# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
+# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
+# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
+# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
+# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
+# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
+# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
+# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
+# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
+# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
+# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
+# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
+# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
+# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
+# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
+# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
+# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
+# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
+# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
+# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
+# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
+# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
+# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
+# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
+# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
+# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
+# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
+# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
+# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
+# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
+# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
+# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
+# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
+# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
+# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
+# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
+# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
+# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
+# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
+# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
+# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
+# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
+# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
+# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
+# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
+# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
+# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
+# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
+# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
+# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
+# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
+# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
+# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
+# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
+# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
+# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
+# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
+# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
+# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
+# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
+# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
+# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
+# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
+# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
+# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
+# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
+# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
+# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
+# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
+# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
+# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
+# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
+# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
+# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
+# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
+# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
+# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
+# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
+# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
+# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
+# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
+# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
+# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
+# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
+# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
+# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
+# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
+# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
+# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
+# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
+# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
+# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
+# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
+# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
+# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
+# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
+# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
+# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
+# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
+# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
+# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
+# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
+# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
+# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
+# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
+# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
+# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
+# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
+# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
+# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
+# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
+# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
+# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
+# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
+# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
+# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
+# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
+# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
+# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
+# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
+# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
+# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
+# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
+# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
+# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
+# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
+# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
+# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
+# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
+# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
+# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
+# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
+# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
+# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
+# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
+# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
+# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
+# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
+# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
+# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
+# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
+# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
+# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
+# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
+# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
+# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
+# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
+# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
+# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
+# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
+# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
+# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
+# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
+# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
+# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
+# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
+# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
+# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
+# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
+# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
+# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
+# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
+# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
+# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
+# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
+# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
+# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
+# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
+# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
+# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
+# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
+# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
+# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
+# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
+# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
+# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
+# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
+# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
+# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
+# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
+# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
+# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
+# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
+# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
+# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
+# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
+# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
+# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
+# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
+# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
+# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
+# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
+# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
+# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
+# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
+# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
+# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
+# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
+# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
+# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
+# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
+# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
+# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
+# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
+# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
+# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
+# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
+# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
+# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
+# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
+# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
+# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
+# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
+# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
+# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
+# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
+# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
+# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
+# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
+# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
+# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
+# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
+# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
+# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
+# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
+# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
+# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
+# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
+# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
+# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
+# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
+# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
+# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
+# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
+# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
+# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
+# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
+# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
+# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
+# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
+# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
+# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
+# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
+# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
+# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
+# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
+# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
+# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
+# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
+# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
+# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
+# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
+# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
+# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
+# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
+# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
+# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
+# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
+# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
+# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
+# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
+# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
+# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
+# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
+# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
+# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
+# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
+# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
+# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
+# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
+# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
+# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
+# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
+# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
+# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
+# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
+# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
+# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
+# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
+# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
+# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
+# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
+# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
+# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
+# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
+# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
+# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
+# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
+# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
+# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
+# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
+# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
+# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
+# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
+# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
+# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
+# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
+# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
+# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
+# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
+# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
+# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
+# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
+# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
+# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
+# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
+# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
+# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
+# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
+# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
+# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
+# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
+# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
+# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
+# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
+# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
+# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
+# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
+# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
+# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
+# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
+# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
+# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
+# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
+# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
+# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
+# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
+# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
+# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
+# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
+# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
+# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
+# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
+# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
+# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
+# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
+# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
+# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
+# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
+# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
+# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
+# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
+# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
+# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
+# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
+# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
+# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
+# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
+# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
+# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
+# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
+# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
+# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
+# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
+# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
+# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
+# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
+# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
+# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
+# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
+# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
+# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
+# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
+# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
+# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
+# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
+# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
+# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
+# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
+# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
+# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
+# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
+# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
+# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
+# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
+# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
+# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
+# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
+# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
+# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
+# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
+# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
+# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
+# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
+# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
+# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
+# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
+# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
+# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
+# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
+# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
+# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
+# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
+# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
+# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
+# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
+# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
+# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
+# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
+# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
+# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
+# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
+# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
+# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
+# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
+# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
+# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
+# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
+# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
+# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
+# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
+# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
+# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
+# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
+# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
+# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
+# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
+# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
+# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
+# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
+# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
+# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
+# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
+# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
+# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
+# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
+# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
+# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
+# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
+# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
+# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
+# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
+# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
+# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
+# 11:38:37 --- INFO: Training process end.
+# 11:38:37 --- INFO: Evaluating process start.
+# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
+# 11:38:37 --- INFO: Evaluating process end.
+# 11:38:37 --- INFO: Predicting process start.
+# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
+# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
+# 11:38:38 --- INFO: Predicting process end.
+# 11:38:38 --- INFO: --------------------- GNN end ----------------------
+# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
+# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
+# 11:38:38 --- INFO: Loading dataset.
+# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:38:39 --- INFO: Calculating NTScore for each niche.
+# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
+# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
+# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
+# 11:38:39 --- INFO: Output NTScore tables.
+# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
16.1.3 visualization
16.1.3.1 load ONTraC results
-
+
The NTScore and binarized niche cluster info were stored in cell metadata
-
-# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
-# <char> <char> <char> <char> <char> <char> <char> <int> <num>
-# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
-# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
-# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
-# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
-# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
-# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
+
+# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
+# <char> <char> <char> <char> <char> <char> <char> <int> <num>
+# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
+# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
+# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
+# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
+# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
+# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477
The probability matrix of each cell assigned to each niche cluster and
connectivity between niche cluster were stored here.
-
-
+
+
16.1.3.2 niche cluster prob
-spatFeatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'niche cluster',
- expression_values = "prob",
- group_by = 'list_ID',
- feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
- point_border_col = 'gray'
- )
+spatFeatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'niche cluster',
+ expression_values = "prob",
+ group_by = 'list_ID',
+ feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
+ point_border_col = 'gray'
+ )
16.1.3.3 binarized niche cluster
-spatPlot2D(giotto_obj,
- spat_unit = "cell",
- group_by = 'list_ID',
- cell_color = "NicheCluster",
- color_as_factor = T,
- point_size = 1,
- point_border_stroke = NA,
- title="Niche cluster")
+spatPlot2D(giotto_obj,
+ spat_unit = "cell",
+ group_by = 'list_ID',
+ cell_color = "NicheCluster",
+ color_as_factor = T,
+ point_size = 1,
+ point_border_stroke = NA,
+ title="Niche cluster")
16.1.3.5 NT score
-spatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'rna',
- group_by = 'list_ID',
- cell_color = "NTScore",
- color_as_factor = FALSE,
- cell_color_gradient = "turbo",
- point_size = 1,
- point_border_stroke = NA
- )
+spatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ group_by = 'list_ID',
+ cell_color = "NTScore",
+ color_as_factor = FALSE,
+ cell_color_gradient = "turbo",
+ point_size = 1,
+ point_border_stroke = NA
+ )
-
+
diff --git a/interoperability-with-other-frameworks.html b/interoperability-with-other-frameworks.html
index 7050533..347d3af 100644
--- a/interoperability-with-other-frameworks.html
+++ b/interoperability-with-other-frameworks.html
@@ -521,10 +521,10 @@ 15 Interoperability with other fr
15.1 Load Giotto object
To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process:
-
+
Load a Giotto mini Visium object, which will be used for demonstrating interoperability.
-
+
15.2 Seurat
@@ -533,99 +533,99 @@ 15.2 Seurat
15.2.1 Conversion of Giotto Object to Seurat Object
To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function
-
+
Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data.
-gToS <- giottoToSeuratV5(gobject = gobject,
- spat_unit = "cell")
-
-plot1 <- VlnPlot(gToS,
- features = "nCount_rna",
- pt.size = 0.1) +
- NoLegend()
-
-plot2 <- SpatialFeaturePlot(gToS,
- features = "nCount_rna",
- pt.size.factor = 2) +
- theme(legend.position = "right")
-
-wrap_plots(plot1, plot2)
+gToS <- giottoToSeuratV5(gobject = gobject,
+ spat_unit = "cell")
+
+plot1 <- VlnPlot(gToS,
+ features = "nCount_rna",
+ pt.size = 0.1) +
+ NoLegend()
+
+plot2 <- SpatialFeaturePlot(gToS,
+ features = "nCount_rna",
+ pt.size.factor = 2) +
+ theme(legend.position = "right")
+
+wrap_plots(plot1, plot2)
15.2.1.1 Apply SCTransform
We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function.
-
+
15.2.1.2 Dimension Reduction
We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object:
-
+
15.2.2 Conversion of Seurat object Back to Giotto Object
To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks.
-giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
- spatial_assay = "rna",
- dim_reduction = c("pca", "umap"),
- sp_network = "Delaunay_network",
- nn_network = c("sNN.pca", "custom_NN" ))
+giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
+ spatial_assay = "rna",
+ dim_reduction = c("pca", "umap"),
+ sp_network = "Delaunay_network",
+ nn_network = c("sNN.pca", "custom_NN" ))
15.2.2.1 Clustering and Plotting UMAP
Here we perform K-means clustering on the UMAP results obtained from the Seurat object:
-## k-means clustering
-giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
- dim_reduction_to_use = "pca")
-
-#Plot UMAP post-clustering to visualize kmeans
-graph2 <- Giotto::plotUMAP(
- gobject = giottoFromSeurat,
- cell_color = "kmeans",
- show_NN_network = TRUE,
- point_size = 2.5
-)
+## k-means clustering
+giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
+ dim_reduction_to_use = "pca")
+
+#Plot UMAP post-clustering to visualize kmeans
+graph2 <- Giotto::plotUMAP(
+ gobject = giottoFromSeurat,
+ cell_color = "kmeans",
+ show_NN_network = TRUE,
+ point_size = 2.5
+)
15.2.2.2 Spatial CoExpression
We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function:
-ranktest <- binSpect(giottoFromSeurat,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "Delaunay_network")
-
-ext_spatial_genes <- ranktest[1:300,]$feats
-
-spat_cor_netw_DT <- detectSpatialCorFeats(
- giottoFromSeurat,
- method = "network",
- spatial_network_name = "Delaunay_network",
- subset_feats = ext_spatial_genes)
-
-top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
- feats = "Dsp",
- show_top_feats = 10)
-
-spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
- name = "spat_netw_clus",
- k = 7)
-heatmSpatialCorFeats(
- giottoFromSeurat,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL),
- save_plot = TRUE,
- show_plot = TRUE,
- return_plot = FALSE,
- save_param = list(base_height = 6, base_width = 8, units = 'cm'))
+ranktest <- binSpect(giottoFromSeurat,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "Delaunay_network")
+
+ext_spatial_genes <- ranktest[1:300,]$feats
+
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ giottoFromSeurat,
+ method = "network",
+ spatial_network_name = "Delaunay_network",
+ subset_feats = ext_spatial_genes)
+
+top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
+ feats = "Dsp",
+ show_top_feats = 10)
+
+spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
+ name = "spat_netw_clus",
+ k = 7)
+heatmSpatialCorFeats(
+ giottoFromSeurat,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ return_plot = FALSE,
+ save_param = list(base_height = 6, base_width = 8, units = 'cm'))
@@ -635,60 +635,60 @@ 15.3 SpatialExperiment
To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries.
-library(SpatialExperiment)
-library(ggspavis)
-library(pheatmap)
-library(scater)
-library(scran)
-library(nnSVG)
+library(SpatialExperiment)
+library(ggspavis)
+library(pheatmap)
+library(scater)
+library(scran)
+library(nnSVG)
15.3.1 Convert Giotto Object to SpatialExperiment Object
To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function.
-
+
The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use:
-
+
15.3.1.1 Identify top spatially variable genes with nnSVG
We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate:
-# One of the assays should be "logcounts"
-# We rename the normalized assay to "logcounts"
-assayNames(spe)[[2]] <- "logcounts"
-
-# Create model matrix for leiden clustering labels
-X <- model.matrix(~ colData(spe)$leiden_clus)
-dim(X)
+# One of the assays should be "logcounts"
+# We rename the normalized assay to "logcounts"
+assayNames(spe)[[2]] <- "logcounts"
+
+# Create model matrix for leiden clustering labels
+X <- model.matrix(~ colData(spe)$leiden_clus)
+dim(X)
- Run nnSVG
This step will take several minutes to run
-
+
15.3.2 Conversion of SpatialExperiment object back to Giotto
We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object.
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-print(giottoFromSPE)
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+print(giottoFromSPE)
15.3.2.1 Plotting top genes from nnSVG in Giotto
Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object.
-ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
- output = "data.table")
-
-ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
-spatFeatPlot2D(giottoFromSPE,
- expression_values = "scaled_rna_cell",
- feats = ext_spatial_genes[1:4],
- point_size = 2)
+ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
+ output = "data.table")
+
+ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID
+spatFeatPlot2D(giottoFromSPE,
+ expression_values = "scaled_rna_cell",
+ feats = ext_spatial_genes[1:4],
+ point_size = 2)
@@ -701,97 +701,97 @@ 15.4 AnnData
15.4.1 Load Required Libraries
To start, we need to load the necessary libraries, including reticulate for interfacing with Python.
-
+
15.4.2 Specify Path for Results
First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions.
-# Specify path to which results may be saved
-results_directory <- "results/03_session4/giotto_anndata_conversion/"
-
-instrs <- showGiottoInstructions(gobject)
-
-mini_gobject <- replaceGiottoInstructions(gobject = gobject,
- instructions = instrs)
+# Specify path to which results may be saved
+results_directory <- "results/03_session4/giotto_anndata_conversion/"
+
+instrs <- showGiottoInstructions(gobject)
+
+mini_gobject <- replaceGiottoInstructions(gobject = gobject,
+ instructions = instrs)
15.4.2.1 Create Default kNN Network
We will create a k-nearest neighbor (kNN) network using mostly default parameters.
-
+
15.4.3 Giotto To AnnData
To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData()
-
+
Next, we import scanpy and perform a series of preprocessing steps on the AnnData object.
-scanpy <- import("scanpy")
-
-adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
-
-# Normalize total counts per cell
-scanpy$pp$normalize_total(adata, target_sum=1e4)
-
-# Log-transform the data
-scanpy$pp$log1p(adata)
-
-# Perform PCA
-scanpy$pp$pca(adata, n_comps=40L)
-
-# Compute the neighborhood graph
-scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
-
-# Run UMAP
-scanpy$tl$umap(adata)
-
-# Save the processed AnnData object
-adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
-
-processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
+scanpy <- import("scanpy")
+
+adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
+
+# Normalize total counts per cell
+scanpy$pp$normalize_total(adata, target_sum=1e4)
+
+# Log-transform the data
+scanpy$pp$log1p(adata)
+
+# Perform PCA
+scanpy$pp$pca(adata, n_comps=40L)
+
+# Compute the neighborhood graph
+scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
+
+# Run UMAP
+scanpy$tl$umap(adata)
+
+# Save the processed AnnData object
+adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
+
+processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"
15.4.4 Convert AnnData to Giotto
Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto.
-
+
15.4.4.1 UMAP Visualization
Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object.
-Giotto::plotUMAP(
- gobject = giottoFromAnndata,
- dim_reduction_name = "umap.ad",
- cell_color = "leiden_clus",
- point_size = 3
-)
+Giotto::plotUMAP(
+ gobject = giottoFromAnndata,
+ dim_reduction_name = "umap.ad",
+ cell_color = "leiden_clus",
+ point_size = 3
+)
subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
- spat_unit = ":all:",
- x_min = 4000,
- x_max = 7000,
- y_min = -6500,
- y_max = -3500,
- z_max = NULL,
- z_min = NULL)
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 2,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_vis_spat_crop'))
-
-spatPlot2D(gobject = subcell_v_brain_crop,
- spat_unit = 'stardist_cell',
- cell_color = 'leiden_clus',
- show_image = T,
- point_size = 0.1,
- point_shape = 'no_border',
- background_color = 'black',
- show_legend = TRUE,
- save_plot = TRUE,
- save_param = list(save_name = '03_ses6_subcell_spat_crop'))subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain,
+ spat_unit = ":all:",
+ x_min = 4000,
+ x_max = 7000,
+ y_min = -6500,
+ y_max = -3500,
+ z_max = NULL,
+ z_min = NULL)
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 2,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_vis_spat_crop'))
+
+spatPlot2D(gobject = subcell_v_brain_crop,
+ spat_unit = 'stardist_cell',
+ cell_color = 'leiden_clus',
+ show_image = T,
+ point_size = 0.1,
+ point_shape = 'no_border',
+ background_color = 'black',
+ show_legend = TRUE,
+ save_plot = TRUE,
+ save_param = list(save_name = '03_ses6_subcell_spat_crop'))Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) diff --git a/interoperability-with-isolated-tools.html b/interoperability-with-isolated-tools.html index ceb2d0a..8cca7e6 100644 --- a/interoperability-with-isolated-tools.html +++ b/interoperability-with-isolated-tools.html @@ -526,1521 +526,1521 @@
16.1.1 Dataset downloaddata_path <- "data"
-dir.create(data_path)
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
- destfile = file.path(data_path,"counts.h5ad"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
- destfile = file.path(data_path,"cell_labels.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))
data_path <- "data"
-dir.create(data_path)
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
- destfile = file.path(data_path,"counts.h5ad"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
- destfile = file.path(data_path,"cell_labels.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
-download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
- destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))data_path <- "data"
+dir.create(data_path)
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad",
+ destfile = file.path(data_path,"counts.h5ad"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv",
+ destfile = file.path(data_path,"cell_labels.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv"))
+download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv",
+ destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv"))16.1.2 Create the Giotto object
-library(Giotto)
-library(reticulate)
-
-## Set instructions
-results_folder <- "results"
-dir.create(results_folder)
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-
-## load data
-
-### meta data
-meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-colnames(meta_df)[[1]] <- "cell_ID"
-slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
-slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
-
-### counts matrix
-scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
-ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
-reticulate::source_python(ad2g_path)
-adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
-X <- extract_expression(adata)
-cID <- extract_cell_IDs(adata)
-fID <- extract_feat_IDs(adata)
-rownames(X) <- fID
-colnames(X) <- cID
-slice1_filter <- cID %in% slice1_meta_df$cell_ID
-slice2_filter <- cID %in% slice2_meta_df$cell_ID
-slice1_exp <- X[,slice1_filter]
-slice2_exp <- X[,slice2_filter]
-
-### cell segmentation to cell location
-segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
-segments_df <- rbind(segments_1_df, segments_2_df)
-loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
-loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
-loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
-centr.x <- apply(loc.use[,loc.x],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-centr.y <- apply(loc.use[,loc.y],1,function(x){
- temp <- lapply(x,function(y){
- as.numeric(unlist(strsplit(y,', ')))
- })
- return (median(unname(unlist(temp))))
-})
-spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
-slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
-slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
-
-## giotto obj
-giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
- spatial_locs = slice1_spatial_locs_df,
- cell_metadata = slice1_meta_df)
-giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
- spatial_locs = slice2_spatial_locs_df,
- cell_metadata = slice2_meta_df)
-giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
- gobject_names = c('mouse2_slice229',
- 'mouse2_slice300'), # name for each samples
- join_method = 'z_stack')
-# We skip the processing process here to save time and use the given cell type
-# annotation directly
-
-ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
- cell_type = "subclass",
- output_path = data_path,
- spat_unit = "cell",
- feat_type = "rna",
- verbose = TRUE)# Cell_ID Sample x y Cell_Type
-# <char> <char> <dbl> <dbl> <char>
-# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
-# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
-# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
-# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
-# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
-# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro library(Giotto)
+library(reticulate)
+
+## Set instructions
+results_folder <- "results"
+dir.create(results_folder)
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+
+## load data
+
+### meta data
+meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+colnames(meta_df)[[1]] <- "cell_ID"
+slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info
+slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info
+
+### counts matrix
+scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env")
+ad2g_path <- system.file("python", "ad2g.py", package = "Giotto")
+reticulate::source_python(ad2g_path)
+adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad"))
+X <- extract_expression(adata)
+cID <- extract_cell_IDs(adata)
+fID <- extract_feat_IDs(adata)
+rownames(X) <- fID
+colnames(X) <- cID
+slice1_filter <- cID %in% slice1_meta_df$cell_ID
+slice2_filter <- cID %in% slice2_meta_df$cell_ID
+slice1_exp <- X[,slice1_filter]
+slice2_exp <- X[,slice2_filter]
+
+### cell segmentation to cell location
+segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers
+segments_df <- rbind(segments_1_df, segments_2_df)
+loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),]
+loc.x <- grep('boundaryX_',colnames(loc.use),value = T)
+loc.y <- grep('boundaryY_',colnames(loc.use),value = T)
+centr.x <- apply(loc.use[,loc.x],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+centr.y <- apply(loc.use[,loc.y],1,function(x){
+ temp <- lapply(x,function(y){
+ as.numeric(unlist(strsplit(y,', ')))
+ })
+ return (median(unname(unlist(temp))))
+})
+spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y)
+slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,]
+slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,]
+
+## giotto obj
+giotto_obj_1 <- createGiottoObject(expression = slice1_exp,
+ spatial_locs = slice1_spatial_locs_df,
+ cell_metadata = slice1_meta_df)
+giotto_obj_2 <- createGiottoObject(expression = slice2_exp,
+ spatial_locs = slice2_spatial_locs_df,
+ cell_metadata = slice2_meta_df)
+giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2),
+ gobject_names = c('mouse2_slice229',
+ 'mouse2_slice300'), # name for each samples
+ join_method = 'z_stack')
+# We skip the processing process here to save time and use the given cell type
+# annotation directly
+
+ONTraC_input <- getONTraCv1Input(gobject = giotto_obj,
+ cell_type = "subclass",
+ output_path = data_path,
+ spat_unit = "cell",
+ feat_type = "rna",
+ verbose = TRUE)# Cell_ID Sample x y Cell_Type
+# <char> <char> <dbl> <dbl> <char>
+# mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT
+# mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC
+# mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT
+# mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo
+# mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT
+# mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions
-spatPlot2D(giotto_obj,
- group_by = "slice_id",
- cell_color = "subclass",
- point_size = 1,
- point_border_stroke = NA,
- legend_text = 6)spatPlot2D(giotto_obj,
+ group_by = "slice_id",
+ cell_color = "subclass",
+ point_size = 1,
+ point_border_stroke = NA,
+ legend_text = 6)
16.1.2.1 Installation ONTraC
You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time.
-source ~/.bash_profile
-conda create -y -n ONTraC python=3.11
-conda activate ONTraC
-pip install git+https://github.com/gyuanlab/ONTraC.git@V2# Retrieving notices: ...working... done
-# Channels:
-# - conda-forge
-# - bioconda
-# - r
-# - pytorch
-# - defaults
-# Platform: osx-arm64
-# Collecting package metadata (repodata.json): ...working... done
-# Solving environment: ...working... done
-#
-# ## Package Plan ##
-#
-# environment location: /Users/wwang/miniconda3/envs/ONTraC
-#
-# added / updated specs:
-# - python=3.11
-#
-#
-# The following packages will be downloaded:
-#
-# package | build
-# ---------------------------|-----------------
-# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
-# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
-# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
-# ------------------------------------------------------------
-# Total: 4.3 MB
-#
-# The following NEW packages will be INSTALLED:
-#
-# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
-# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
-# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
-# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
-# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
-# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
-# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
-# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
-# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
-# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
-# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
-# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
-# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
-# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
-# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
-# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
-#
-#
-#
-# Downloading and Extracting Packages: ...working... done
-# Preparing transaction: ...working... done
-# Verifying transaction: ...working... done
-# Executing transaction: ...working... done
-# #
-# # To activate this environment, use
-# #
-# # $ conda activate ONTraC
-# #
-# # To deactivate an active environment, use
-# #
-# # $ conda deactivate
-#
-# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
-# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
-# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
-# Running command git checkout -b V2 --track origin/V2
-# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
-# Installing build dependencies: started
-# Switched to a new branch 'V2'
-# branch 'V2' set up to track 'origin/V2'.
-# Installing build dependencies: finished with status 'done'
-# Getting requirements to build wheel: started
-# Getting requirements to build wheel: finished with status 'done'
-# Preparing metadata (pyproject.toml): started
-# Preparing metadata (pyproject.toml): finished with status 'done'
-# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
-# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
-# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
-# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
-# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
-# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
-# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
-# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
-# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
-# Collecting session-info (from ONTraC==2.0a9.dev7)
-# Using cached session_info-1.0.0-py3-none-any.whl
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
-# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
-# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
-# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
-# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
-# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
-# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
-# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
-# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
-# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
-# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
-# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
-# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
-# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
-# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
-# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
-# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
-# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
-# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
-# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
-# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
-# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
-# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
-# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
-# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
-# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
-# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
-# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
-# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
-# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
-# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
-# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
-# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
-# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
-# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
-# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
-# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
-# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
-# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
-# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
-# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
-# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
-# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
-# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
-# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
-# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
-# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
-# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
-# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
-# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
-# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
-# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
-# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
-# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
-# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
-# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
-# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
-# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
-# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
-# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
-# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
-# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
-# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
-# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
-# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
-# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
-# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
-# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
-# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
-# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
-# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
-# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
-# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
-# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
-# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
-# Using cached idna-3.7-py3-none-any.whl (66 kB)
-# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
-# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
-# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
-# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
-# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
-# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
-# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
-# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
-# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
-# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
-# Building wheels for collected packages: ONTraC
-# Building wheel for ONTraC (pyproject.toml): started
-# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
-# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
-# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
-# Successfully built ONTraC
-# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
-# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4source ~/.bash_profile
+conda create -y -n ONTraC python=3.11
+conda activate ONTraC
+pip install git+https://github.com/gyuanlab/ONTraC.git@V2# Retrieving notices: ...working... done
+# Channels:
+# - conda-forge
+# - bioconda
+# - r
+# - pytorch
+# - defaults
+# Platform: osx-arm64
+# Collecting package metadata (repodata.json): ...working... done
+# Solving environment: ...working... done
+#
+# ## Package Plan ##
+#
+# environment location: /Users/wwang/miniconda3/envs/ONTraC
+#
+# added / updated specs:
+# - python=3.11
+#
+#
+# The following packages will be downloaded:
+#
+# package | build
+# ---------------------------|-----------------
+# bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge
+# openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge
+# setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge
+# ------------------------------------------------------------
+# Total: 4.3 MB
+#
+# The following NEW packages will be INSTALLED:
+#
+# bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7
+# ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0
+# libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0
+# libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5
+# libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0
+# libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1
+# ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0
+# openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2
+# pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0
+# python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython
+# readline conda-forge/osx-arm64::readline-8.2-h92ec313_1
+# setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0
+# tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1
+# tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0
+# wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1
+# xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0
+#
+#
+#
+# Downloading and Extracting Packages: ...working... done
+# Preparing transaction: ...working... done
+# Verifying transaction: ...working... done
+# Executing transaction: ...working... done
+# #
+# # To activate this environment, use
+# #
+# # $ conda activate ONTraC
+# #
+# # To deactivate an active environment, use
+# #
+# # $ conda deactivate
+#
+# Collecting git+https://github.com/gyuanlab/ONTraC.git@V2
+# Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3
+# Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3
+# Running command git checkout -b V2 --track origin/V2
+# Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d
+# Installing build dependencies: started
+# Switched to a new branch 'V2'
+# branch 'V2' set up to track 'origin/V2'.
+# Installing build dependencies: finished with status 'done'
+# Getting requirements to build wheel: started
+# Getting requirements to build wheel: finished with status 'done'
+# Preparing metadata (pyproject.toml): started
+# Preparing metadata (pyproject.toml): finished with status 'done'
+# Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB)
+# Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB)
+# Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB)
+# Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB)
+# Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7)
+# Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB)
+# Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7)
+# Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB)
+# Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB)
+# Collecting session-info (from ONTraC==2.0a9.dev7)
+# Using cached session_info-1.0.0-py3-none-any.whl
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00
+# Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB)
+# Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB)
+# Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB)
+# Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB)
+# Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB)
+# Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB)
+# Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB)
+# Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB)
+# Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB)
+# Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB)
+# Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB)
+# Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB)
+# Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB)
+# Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB)
+# Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB)
+# Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB)
+# Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB)
+# Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB)
+# Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB)
+# Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB)
+# Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB)
+# Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB)
+# Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB)
+# Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7)
+# Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB)
+# Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB)
+# Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB)
+# Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB)
+# Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB)
+# Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB)
+# Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB)
+# Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB)
+# Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB)
+# Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB)
+# Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB)
+# Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB)
+# Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7)
+# Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB)
+# Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7)
+# Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB)
+# Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB)
+# Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB)
+# Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB)
+# Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB)
+# Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB)
+# Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB)
+# Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB)
+# Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB)
+# Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB)
+# Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB)
+# Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB)
+# Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB)
+# Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB)
+# Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB)
+# Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB)
+# Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB)
+# Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB)
+# Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB)
+# Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB)
+# Using cached filelock-3.15.4-py3-none-any.whl (16 kB)
+# Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB)
+# Using cached jinja2-3.1.4-py3-none-any.whl (133 kB)
+# Using cached networkx-3.3-py3-none-any.whl (1.7 MB)
+# Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB)
+# Using cached requests-2.32.3-py3-none-any.whl (64 kB)
+# Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB)
+# Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB)
+# ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00
+# Using cached tqdm-4.66.4-py3-none-any.whl (78 kB)
+# Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB)
+# Using cached attrs-23.2.0-py3-none-any.whl (60 kB)
+# Using cached certifi-2024.7.4-py3-none-any.whl (162 kB)
+# Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB)
+# Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB)
+# Using cached idna-3.7-py3-none-any.whl (66 kB)
+# Using cached joblib-1.4.2-py3-none-any.whl (301 kB)
+# Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB)
+# Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB)
+# Using cached mpmath-1.3.0-py3-none-any.whl (536 kB)
+# Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB)
+# Using cached six-1.16.0-py2.py3-none-any.whl (11 kB)
+# Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB)
+# Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB)
+# Using cached urllib3-2.2.2-py3-none-any.whl (121 kB)
+# Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB)
+# Building wheels for collected packages: ONTraC
+# Building wheel for ONTraC (pyproject.toml): started
+# Building wheel for ONTraC (pyproject.toml): finished with status 'done'
+# Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179
+# Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a
+# Successfully built ONTraC
+# Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC
+# Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.416.1.2.2 Running ONTraC
-
-source ~/.bash_profile
-conda activate ONTraC_v2_py311
-ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log# ##################################################################################
-#
-# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
-# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
-# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
-# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
-# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
-#
-# version: 2.0a11
-#
-# ##################################################################################
-# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
-# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
-# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
-# 11:33:23 --- INFO: -------- I/O options -------
-# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
-# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
-# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
-# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
-# 11:33:23 --- INFO: -------- preprocessing options -------
-# 11:33:23 --- INFO: resolution: 10.0
-# 11:33:23 --- INFO: -------- niche net constr options -------
-# 11:33:23 --- INFO: n_cpu: 4
-# 11:33:23 --- INFO: n_neighbors: 50
-# 11:33:23 --- INFO: n_local: 20
-# 11:33:23 --- INFO: embedding_adjust: False
-# 11:33:23 --- INFO: sigma: 1
-# 11:33:23 --- INFO: -------- train options -------
-# 11:33:23 --- INFO: device: cpu
-# 11:33:23 --- INFO: epochs: 1000
-# 11:33:23 --- INFO: batch_size: 0
-# 11:33:23 --- INFO: patience: 100
-# 11:33:23 --- INFO: min_delta: 0.001
-# 11:33:23 --- INFO: min_epochs: 50
-# 11:33:23 --- INFO: seed: 42
-# 11:33:23 --- INFO: lr: 0.03
-# 11:33:23 --- INFO: hidden_feats: 4
-# 11:33:23 --- INFO: k: 6
-# 11:33:23 --- INFO: modularity_loss_weight: 1.0
-# 11:33:23 --- INFO: purity_loss_weight: 100.0
-# 11:33:23 --- INFO: regularization_loss_weight: 0.1
-# 11:33:23 --- INFO: beta: 0.3
-# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
-# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
-# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
-# 11:33:23 --- INFO: ------------- Niche network construct ---------------
-# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
-# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
-# 11:33:28 --- INFO: Generating samples.yaml file.
-# 11:33:28 --- INFO: ------------ Niche network construct end ------------
-# 11:33:28 --- INFO: ------------------------ GNN ------------------------
-# 11:33:28 --- INFO: Loading dataset.
-# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
-# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
-# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
-# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
-# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
-# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
-# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
-# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
-# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
-# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
-# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
-# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
-# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
-# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
-# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
-# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
-# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
-# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
-# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
-# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
-# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
-# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
-# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
-# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
-# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
-# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
-# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
-# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
-# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
-# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
-# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
-# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
-# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
-# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
-# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
-# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
-# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
-# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
-# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
-# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
-# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
-# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
-# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
-# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
-# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
-# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
-# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
-# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
-# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
-# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
-# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
-# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
-# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
-# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
-# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
-# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
-# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
-# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
-# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
-# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
-# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
-# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
-# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
-# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
-# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
-# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
-# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
-# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
-# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
-# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
-# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
-# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
-# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
-# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
-# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
-# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
-# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
-# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
-# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
-# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
-# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
-# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
-# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
-# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
-# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
-# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
-# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
-# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
-# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
-# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
-# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
-# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
-# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
-# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
-# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
-# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
-# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
-# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
-# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
-# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
-# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
-# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
-# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
-# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
-# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
-# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
-# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
-# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
-# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
-# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
-# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
-# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
-# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
-# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
-# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
-# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
-# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
-# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
-# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
-# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
-# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
-# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
-# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
-# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
-# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
-# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
-# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
-# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
-# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
-# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
-# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
-# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
-# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
-# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
-# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
-# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
-# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
-# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
-# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
-# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
-# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
-# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
-# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
-# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
-# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
-# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
-# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
-# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
-# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
-# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
-# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
-# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
-# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
-# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
-# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
-# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
-# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
-# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
-# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
-# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
-# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
-# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
-# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
-# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
-# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
-# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
-# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
-# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
-# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
-# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
-# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
-# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
-# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
-# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
-# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
-# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
-# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
-# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
-# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
-# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
-# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
-# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
-# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
-# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
-# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
-# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
-# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
-# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
-# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
-# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
-# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
-# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
-# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
-# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
-# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
-# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
-# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
-# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
-# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
-# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
-# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
-# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
-# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
-# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
-# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
-# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
-# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
-# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
-# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
-# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
-# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
-# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
-# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
-# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
-# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
-# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
-# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
-# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
-# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
-# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
-# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
-# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
-# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
-# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
-# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
-# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
-# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
-# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
-# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
-# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
-# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
-# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
-# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
-# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
-# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
-# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
-# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
-# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
-# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
-# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
-# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
-# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
-# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
-# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
-# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
-# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
-# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
-# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
-# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
-# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
-# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
-# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
-# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
-# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
-# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
-# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
-# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
-# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
-# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
-# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
-# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
-# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
-# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
-# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
-# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
-# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
-# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
-# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
-# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
-# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
-# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
-# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
-# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
-# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
-# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
-# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
-# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
-# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
-# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
-# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
-# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
-# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
-# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
-# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
-# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
-# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
-# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
-# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
-# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
-# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
-# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
-# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
-# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
-# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
-# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
-# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
-# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
-# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
-# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
-# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
-# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
-# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
-# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
-# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
-# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
-# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
-# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
-# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
-# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
-# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
-# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
-# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
-# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
-# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
-# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
-# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
-# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
-# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
-# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
-# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
-# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
-# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
-# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
-# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
-# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
-# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
-# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
-# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
-# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
-# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
-# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
-# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
-# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
-# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
-# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
-# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
-# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
-# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
-# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
-# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
-# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
-# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
-# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
-# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
-# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
-# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
-# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
-# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
-# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
-# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
-# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
-# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
-# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
-# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
-# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
-# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
-# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
-# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
-# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
-# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
-# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
-# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
-# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
-# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
-# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
-# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
-# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
-# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
-# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
-# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
-# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
-# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
-# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
-# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
-# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
-# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
-# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
-# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
-# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
-# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
-# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
-# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
-# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
-# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
-# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
-# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
-# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
-# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
-# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
-# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
-# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
-# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
-# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
-# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
-# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
-# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
-# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
-# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
-# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
-# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
-# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
-# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
-# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
-# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
-# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
-# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
-# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
-# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
-# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
-# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
-# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
-# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
-# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
-# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
-# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
-# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
-# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
-# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
-# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
-# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
-# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
-# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
-# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
-# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
-# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
-# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
-# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
-# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
-# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
-# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
-# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
-# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
-# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
-# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
-# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
-# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
-# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
-# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
-# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
-# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
-# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
-# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
-# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
-# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
-# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
-# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
-# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
-# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
-# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
-# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
-# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
-# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
-# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
-# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
-# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
-# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
-# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
-# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
-# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
-# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
-# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
-# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
-# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
-# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
-# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
-# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
-# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
-# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
-# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
-# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
-# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
-# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
-# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
-# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
-# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
-# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
-# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
-# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
-# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
-# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
-# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
-# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
-# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
-# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
-# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
-# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
-# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
-# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
-# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
-# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
-# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
-# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
-# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
-# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
-# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
-# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
-# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
-# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
-# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
-# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
-# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
-# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
-# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
-# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
-# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
-# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
-# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
-# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
-# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
-# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
-# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
-# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
-# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
-# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
-# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
-# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
-# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
-# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
-# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
-# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
-# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
-# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
-# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
-# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
-# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
-# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
-# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
-# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
-# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
-# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
-# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
-# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
-# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
-# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
-# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
-# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
-# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
-# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
-# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
-# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
-# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
-# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
-# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
-# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
-# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
-# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
-# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
-# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
-# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
-# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
-# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
-# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
-# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
-# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
-# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
-# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
-# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
-# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
-# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
-# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
-# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
-# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
-# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
-# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
-# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
-# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
-# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
-# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
-# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
-# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
-# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
-# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
-# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
-# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
-# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
-# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
-# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
-# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
-# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
-# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
-# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
-# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
-# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
-# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
-# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
-# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
-# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
-# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
-# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
-# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
-# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
-# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
-# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
-# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
-# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
-# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
-# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
-# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
-# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
-# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
-# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
-# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
-# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
-# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
-# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
-# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
-# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
-# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
-# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
-# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
-# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
-# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
-# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
-# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
-# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
-# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
-# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
-# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
-# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
-# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
-# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
-# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
-# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
-# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
-# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
-# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
-# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
-# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
-# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
-# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
-# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
-# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
-# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
-# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
-# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
-# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
-# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
-# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
-# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
-# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
-# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
-# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
-# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
-# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
-# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
-# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
-# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
-# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
-# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
-# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
-# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
-# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
-# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
-# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
-# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
-# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
-# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
-# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
-# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
-# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
-# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
-# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
-# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
-# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
-# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
-# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
-# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
-# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
-# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
-# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
-# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
-# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
-# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
-# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
-# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
-# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
-# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
-# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
-# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
-# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
-# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
-# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
-# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
-# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
-# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
-# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
-# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
-# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
-# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
-# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
-# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
-# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
-# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
-# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
-# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
-# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
-# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
-# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
-# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
-# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
-# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
-# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
-# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
-# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
-# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
-# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
-# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
-# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
-# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
-# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
-# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
-# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
-# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
-# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
-# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
-# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
-# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
-# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
-# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
-# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
-# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
-# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
-# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
-# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
-# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
-# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
-# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
-# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
-# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
-# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
-# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
-# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
-# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
-# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
-# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
-# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
-# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
-# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
-# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
-# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
-# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
-# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
-# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
-# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
-# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
-# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
-# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
-# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
-# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
-# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
-# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
-# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
-# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
-# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
-# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
-# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
-# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
-# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
-# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
-# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
-# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
-# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
-# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
-# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
-# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
-# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
-# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
-# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
-# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
-# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
-# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
-# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
-# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
-# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
-# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
-# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
-# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
-# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
-# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
-# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
-# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
-# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
-# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
-# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
-# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
-# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
-# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
-# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
-# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
-# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
-# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
-# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
-# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
-# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
-# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
-# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
-# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
-# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
-# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
-# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
-# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
-# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
-# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
-# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
-# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
-# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
-# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
-# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
-# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
-# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
-# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
-# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
-# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
-# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
-# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
-# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
-# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
-# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
-# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
-# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
-# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
-# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
-# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
-# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
-# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
-# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
-# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
-# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
-# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
-# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
-# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
-# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
-# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
-# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
-# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
-# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
-# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
-# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
-# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
-# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
-# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
-# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
-# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
-# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
-# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
-# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
-# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
-# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
-# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
-# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
-# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
-# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
-# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
-# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
-# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
-# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
-# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
-# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
-# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
-# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
-# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
-# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
-# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
-# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
-# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
-# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
-# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
-# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
-# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
-# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
-# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
-# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
-# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
-# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
-# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
-# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
-# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
-# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
-# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
-# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
-# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
-# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
-# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
-# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
-# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
-# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
-# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
-# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
-# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
-# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
-# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
-# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
-# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
-# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
-# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
-# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
-# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
-# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
-# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
-# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
-# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
-# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
-# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
-# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
-# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
-# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
-# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
-# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
-# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
-# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
-# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
-# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
-# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
-# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
-# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
-# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
-# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
-# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
-# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
-# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
-# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
-# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
-# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
-# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
-# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
-# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
-# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
-# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
-# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
-# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
-# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
-# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
-# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
-# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
-# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
-# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
-# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
-# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
-# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
-# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
-# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
-# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
-# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
-# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
-# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
-# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
-# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
-# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
-# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
-# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
-# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
-# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
-# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
-# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
-# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
-# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
-# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
-# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
-# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
-# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
-# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
-# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
-# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
-# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
-# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
-# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
-# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
-# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
-# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
-# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
-# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
-# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
-# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
-# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
-# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
-# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
-# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
-# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
-# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
-# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
-# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
-# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
-# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
-# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
-# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
-# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
-# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
-# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
-# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
-# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
-# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
-# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
-# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
-# 11:38:37 --- INFO: Training process end.
-# 11:38:37 --- INFO: Evaluating process start.
-# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
-# 11:38:37 --- INFO: Evaluating process end.
-# 11:38:37 --- INFO: Predicting process start.
-# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
-# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
-# 11:38:38 --- INFO: Predicting process end.
-# 11:38:38 --- INFO: --------------------- GNN end ----------------------
-# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
-# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
-# 11:38:38 --- INFO: Loading dataset.
-# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
-# Processing...
-# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# Done!
-# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
-# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
-# 11:38:39 --- INFO: Calculating NTScore for each niche.
-# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
-# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
-# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
-# 11:38:39 --- INFO: Output NTScore tables.
-# 11:38:39 --- INFO: --------------- Niche trajectory end ----------------
+source ~/.bash_profile
+conda activate ONTraC_v2_py311
+ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log# ##################################################################################
+#
+# ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█
+# ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀
+# ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██
+# ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄
+# ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀
+#
+# version: 2.0a11
+#
+# ##################################################################################
+# 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten.
+# 11:33:23 --- WARNING: CUDA is not available, use CPU instead.
+# 11:33:23 --- INFO: ------------------ RUN params memo ------------------
+# 11:33:23 --- INFO: -------- I/O options -------
+# 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv
+# 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir
+# 11:33:23 --- INFO: GNN output directory: data/GNN_dir
+# 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir
+# 11:33:23 --- INFO: -------- preprocessing options -------
+# 11:33:23 --- INFO: resolution: 10.0
+# 11:33:23 --- INFO: -------- niche net constr options -------
+# 11:33:23 --- INFO: n_cpu: 4
+# 11:33:23 --- INFO: n_neighbors: 50
+# 11:33:23 --- INFO: n_local: 20
+# 11:33:23 --- INFO: embedding_adjust: False
+# 11:33:23 --- INFO: sigma: 1
+# 11:33:23 --- INFO: -------- train options -------
+# 11:33:23 --- INFO: device: cpu
+# 11:33:23 --- INFO: epochs: 1000
+# 11:33:23 --- INFO: batch_size: 0
+# 11:33:23 --- INFO: patience: 100
+# 11:33:23 --- INFO: min_delta: 0.001
+# 11:33:23 --- INFO: min_epochs: 50
+# 11:33:23 --- INFO: seed: 42
+# 11:33:23 --- INFO: lr: 0.03
+# 11:33:23 --- INFO: hidden_feats: 4
+# 11:33:23 --- INFO: k: 6
+# 11:33:23 --- INFO: modularity_loss_weight: 1.0
+# 11:33:23 --- INFO: purity_loss_weight: 100.0
+# 11:33:23 --- INFO: regularization_loss_weight: 0.1
+# 11:33:23 --- INFO: beta: 0.3
+# 11:33:23 --- INFO: ---------------- Niche trajectory options ----------------
+# 11:33:23 --- INFO: Equally spaced niche cluster scores: True
+# 11:33:23 --- INFO: --------------- RUN params memo end -----------------
+# 11:33:23 --- INFO: ------------- Niche network construct ---------------
+# 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229.
+# 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300.
+# 11:33:28 --- INFO: Generating samples.yaml file.
+# 11:33:28 --- INFO: ------------ Niche network construct end ------------
+# 11:33:28 --- INFO: ------------------------ GNN ------------------------
+# 11:33:28 --- INFO: Loading dataset.
+# 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295
+# 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042
+# 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327
+# 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115
+# 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213
+# 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614
+# 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403
+# 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237
+# 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202
+# 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215
+# 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228
+# 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795
+# 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693
+# 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767
+# 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586
+# 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281
+# 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438
+# 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432
+# 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744
+# 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989
+# 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189
+# 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243
+# 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619
+# 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271
+# 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676
+# 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364
+# 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671
+# 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218
+# 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395
+# 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057
+# 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591
+# 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211
+# 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326
+# 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168
+# 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018
+# 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847
+# 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625
+# 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743
+# 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154
+# 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737
+# 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406
+# 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176
+# 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143
+# 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268
+# 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245
+# 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431
+# 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931
+# 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054
+# 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113
+# 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119
+# 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123
+# 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827
+# 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282
+# 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788
+# 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721
+# 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239
+# 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681
+# 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721
+# 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827
+# 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585
+# 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262
+# 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794
+# 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467
+# 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049
+# 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771
+# 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307
+# 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605
+# 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451
+# 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646
+# 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451
+# 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704
+# 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258
+# 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778
+# 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272
+# 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008
+# 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067
+# 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049
+# 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457
+# 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127
+# 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987
+# 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915
+# 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015
+# 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767
+# 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737
+# 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466
+# 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513
+# 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802
+# 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072
+# 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461
+# 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236
+# 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504
+# 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481
+# 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282
+# 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388
+# 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072
+# 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827
+# 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992
+# 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055
+# 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778
+# 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925
+# 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407
+# 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309
+# 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949
+# 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067
+# 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663
+# 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325
+# 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474
+# 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726
+# 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705
+# 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333
+# 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843
+# 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393
+# 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619
+# 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496
+# 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177
+# 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263
+# 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973
+# 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945
+# 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254
+# 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053
+# 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514
+# 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541
+# 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965
+# 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392
+# 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478
+# 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168
+# 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469
+# 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927
+# 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999
+# 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427
+# 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604
+# 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314
+# 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694
+# 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767
+# 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556
+# 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943
+# 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053
+# 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258
+# 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704
+# 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198
+# 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036
+# 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676
+# 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661
+# 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786
+# 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285
+# 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848
+# 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128
+# 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801
+# 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814
+# 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846
+# 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953
+# 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254
+# 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184
+# 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286
+# 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667
+# 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423
+# 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705
+# 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502
+# 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898
+# 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392
+# 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678
+# 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298
+# 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229
+# 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657
+# 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547
+# 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473
+# 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443
+# 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958
+# 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592
+# 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924
+# 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203
+# 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846
+# 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558
+# 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373
+# 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416
+# 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485
+# 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494
+# 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866
+# 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513
+# 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346
+# 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425
+# 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465
+# 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227
+# 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359
+# 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986
+# 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145
+# 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233
+# 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022
+# 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185
+# 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097
+# 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677
+# 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471
+# 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368
+# 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055
+# 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637
+# 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069
+# 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291
+# 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572
+# 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808
+# 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703
+# 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254
+# 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093
+# 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355
+# 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058
+# 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682
+# 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127
+# 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444
+# 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834
+# 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612
+# 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012
+# 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844
+# 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186
+# 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351
+# 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285
+# 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035
+# 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017
+# 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856
+# 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863
+# 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868
+# 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535
+# 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841
+# 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179
+# 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073
+# 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764
+# 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268
+# 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765
+# 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823
+# 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526
+# 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059
+# 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852
+# 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169
+# 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171
+# 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011
+# 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057
+# 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762
+# 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224
+# 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306
+# 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008
+# 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728
+# 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967
+# 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112
+# 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982
+# 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989
+# 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769
+# 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553
+# 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711
+# 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987
+# 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575
+# 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671
+# 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935
+# 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848
+# 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713
+# 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088
+# 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323
+# 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125
+# 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462
+# 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306
+# 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993
+# 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898
+# 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551
+# 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307
+# 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731
+# 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807
+# 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386
+# 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498
+# 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248
+# 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011
+# 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683
+# 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789
+# 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182
+# 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042
+# 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909
+# 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095
+# 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123
+# 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813
+# 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983
+# 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065
+# 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414
+# 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968
+# 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667
+# 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242
+# 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797
+# 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512
+# 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495
+# 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133
+# 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234
+# 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307
+# 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548
+# 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122
+# 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461
+# 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314
+# 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405
+# 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359
+# 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939
+# 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219
+# 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558
+# 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166
+# 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447
+# 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818
+# 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643
+# 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463
+# 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347
+# 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546
+# 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402
+# 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362
+# 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412
+# 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984
+# 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751
+# 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184
+# 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229
+# 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754
+# 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276
+# 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677
+# 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853
+# 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818
+# 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556
+# 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632
+# 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225
+# 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286
+# 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143
+# 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827
+# 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028
+# 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162
+# 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371
+# 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775
+# 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895
+# 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702
+# 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224
+# 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508
+# 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688
+# 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619
+# 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526
+# 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604
+# 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793
+# 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123
+# 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128
+# 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152
+# 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356
+# 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011
+# 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219
+# 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119
+# 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218
+# 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143
+# 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087
+# 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602
+# 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637
+# 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906
+# 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044
+# 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261
+# 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937
+# 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504
+# 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906
+# 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666
+# 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948
+# 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098
+# 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097
+# 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423
+# 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002
+# 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576
+# 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759
+# 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349
+# 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333
+# 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914
+# 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072
+# 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213
+# 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981
+# 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031
+# 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903
+# 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462
+# 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171
+# 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778
+# 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147
+# 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935
+# 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394
+# 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912
+# 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443
+# 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619
+# 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256
+# 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849
+# 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531
+# 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978
+# 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491
+# 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723
+# 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325
+# 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344
+# 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808
+# 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569
+# 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734
+# 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553
+# 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281
+# 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143
+# 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331
+# 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752
+# 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396
+# 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693
+# 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532
+# 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463
+# 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985
+# 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612
+# 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851
+# 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363
+# 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112
+# 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577
+# 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204
+# 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084
+# 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888
+# 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332
+# 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625
+# 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215
+# 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133
+# 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281
+# 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625
+# 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593
+# 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073
+# 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599
+# 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005
+# 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351
+# 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308
+# 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055
+# 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298
+# 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368
+# 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323
+# 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068
+# 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432
+# 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604
+# 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621
+# 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729
+# 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945
+# 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894
+# 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742
+# 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422
+# 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878
+# 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868
+# 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922
+# 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472
+# 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894
+# 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289
+# 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311
+# 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819
+# 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058
+# 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106
+# 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717
+# 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448
+# 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344
+# 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296
+# 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515
+# 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563
+# 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645
+# 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425
+# 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367
+# 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059
+# 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324
+# 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529
+# 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318
+# 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486
+# 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419
+# 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465
+# 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329
+# 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988
+# 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622
+# 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401
+# 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472
+# 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636
+# 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743
+# 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169
+# 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689
+# 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962
+# 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523
+# 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413
+# 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537
+# 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263
+# 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638
+# 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737
+# 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818
+# 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456
+# 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389
+# 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816
+# 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964
+# 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295
+# 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897
+# 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181
+# 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889
+# 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882
+# 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247
+# 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695
+# 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932
+# 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462
+# 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585
+# 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213
+# 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854
+# 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196
+# 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762
+# 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372
+# 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428
+# 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258
+# 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656
+# 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776
+# 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752
+# 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958
+# 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211
+# 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449
+# 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236
+# 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107
+# 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957
+# 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474
+# 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239
+# 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778
+# 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226
+# 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228
+# 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198
+# 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854
+# 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806
+# 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431
+# 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188
+# 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269
+# 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315
+# 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883
+# 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135
+# 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026
+# 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599
+# 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296
+# 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107
+# 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035
+# 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363
+# 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629
+# 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176
+# 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155
+# 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865
+# 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037
+# 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078
+# 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882
+# 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671
+# 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294
+# 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854
+# 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187
+# 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931
+# 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968
+# 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687
+# 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495
+# 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275
+# 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167
+# 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776
+# 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459
+# 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771
+# 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593
+# 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551
+# 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837
+# 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927
+# 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806
+# 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178
+# 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342
+# 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986
+# 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375
+# 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588
+# 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253
+# 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324
+# 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195
+# 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474
+# 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226
+# 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533
+# 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303
+# 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206
+# 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095
+# 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031
+# 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639
+# 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158
+# 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273
+# 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255
+# 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058
+# 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907
+# 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681
+# 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035
+# 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988
+# 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094
+# 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773
+# 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846
+# 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755
+# 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396
+# 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126
+# 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214
+# 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078
+# 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509
+# 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715
+# 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316
+# 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964
+# 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904
+# 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367
+# 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277
+# 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069
+# 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883
+# 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008
+# 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823
+# 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825
+# 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375
+# 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581
+# 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237
+# 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057
+# 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983
+# 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237
+# 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745
+# 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535
+# 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628
+# 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906
+# 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785
+# 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097
+# 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409
+# 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364
+# 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753
+# 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977
+# 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827
+# 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662
+# 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198
+# 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182
+# 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508
+# 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552
+# 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626
+# 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582
+# 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523
+# 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001
+# 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239
+# 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001
+# 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374
+# 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178
+# 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566
+# 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042
+# 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012
+# 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356
+# 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565
+# 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729
+# 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222
+# 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863
+# 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041
+# 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716
+# 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298
+# 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323
+# 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472
+# 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283
+# 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159
+# 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846
+# 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383
+# 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009
+# 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934
+# 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143
+# 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755
+# 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247
+# 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381
+# 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336
+# 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575
+# 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694
+# 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888
+# 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515
+# 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843
+# 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753
+# 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111
+# 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215
+# 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917
+# 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812
+# 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006
+# 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887
+# 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208
+# 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722
+# 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318
+# 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735
+# 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123
+# 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705
+# 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273
+# 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596
+# 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288
+# 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094
+# 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214
+# 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422
+# 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541
+# 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123
+# 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316
+# 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675
+# 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212
+# 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704
+# 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793
+# 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644
+# 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166
+# 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118
+# 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657
+# 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685
+# 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299
+# 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715
+# 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297
+# 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926
+# 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568
+# 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374
+# 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732
+# 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776
+# 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925
+# 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283
+# 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909
+# 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373
+# 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044
+# 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716
+# 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134
+# 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894
+# 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582
+# 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164
+# 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955
+# 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732
+# 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523
+# 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506
+# 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224
+# 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627
+# 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014
+# 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581
+# 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073
+# 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625
+# 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267
+# 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193
+# 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312
+# 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924
+# 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476
+# 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192
+# 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058
+# 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625
+# 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163
+# 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118
+# 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237
+# 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267
+# 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043
+# 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176
+# 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087
+# 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446
+# 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655
+# 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177
+# 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998
+# 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849
+# 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251
+# 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878
+# 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057
+# 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445
+# 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356
+# 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625
+# 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192
+# 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655
+# 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088
+# 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669
+# 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758
+# 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176
+# 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534
+# 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728
+# 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325
+# 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728
+# 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012
+# 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668
+# 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265
+# 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728
+# 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833
+# 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164
+# 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326
+# 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908
+# 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251
+# 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893
+# 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625
+# 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192
+# 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267
+# 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849
+# 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894
+# 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267
+# 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311
+# 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042
+# 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326
+# 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251
+# 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012
+# 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386
+# 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939
+# 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461
+# 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655
+# 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267
+# 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699
+# 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386
+# 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655
+# 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655
+# 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461
+# 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043
+# 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774
+# 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043
+# 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968
+# 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774
+# 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864
+# 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148
+# 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461
+# 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491
+# 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804
+# 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819
+# 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058
+# 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043
+# 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073
+# 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656
+# 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088
+# 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491
+# 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804
+# 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968
+# 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505
+# 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356
+# 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506
+# 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625
+# 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148
+# 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656
+# 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073
+# 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043
+# 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864
+# 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387
+# 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685
+# 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073
+# 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178
+# 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447
+# 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686
+# 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999
+# 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088
+# 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208
+# 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596
+# 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894
+# 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701
+# 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088
+# 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134
+# 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462
+# 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193
+# 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536
+# 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208
+# 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088
+# 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909
+# 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103
+# 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282
+# 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387
+# 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462
+# 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029
+# 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403
+# 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955
+# 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284
+# 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312
+# 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403
+# 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491
+# 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208
+# 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372
+# 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761
+# 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388
+# 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925
+# 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238
+# 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894
+# 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282
+# 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582
+# 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776
+# 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075
+# 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344
+# 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105
+# 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508
+# 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673
+# 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911
+# 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715
+# 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598
+# 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554
+# 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659
+# 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763
+# 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614
+# 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241
+# 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554
+# 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897
+# 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405
+# 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584
+# 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808
+# 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823
+# 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854
+# 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062
+# 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958
+# 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551
+# 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018
+# 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555
+# 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775
+# 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049
+# 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243
+# 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212
+# 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178
+# 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407
+# 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243
+# 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033
+# 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377
+# 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899
+# 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362
+# 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601
+# 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483
+# 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005
+# 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005
+# 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417
+# 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662
+# 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387
+# 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348
+# 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631
+# 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631
+# 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438
+# 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402
+# 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676
+# 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826
+# 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498
+# 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722
+# 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185
+# 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349
+# 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812
+# 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275
+# 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857
+# 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379
+# 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887
+# 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574
+# 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798
+# 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044
+# 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626
+# 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917
+# 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574
+# 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215
+# 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783
+# 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529
+# 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082
+# 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172
+# 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351
+# 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948
+# 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888
+# 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649
+# 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291
+# 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918
+# 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784
+# 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903
+# 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366
+# 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059
+# 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247
+# 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366
+# 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829
+# 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471
+# 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128
+# 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442
+# 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089
+# 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492
+# 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622
+# 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815
+# 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696
+# 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104
+# 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457
+# 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159
+# 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547
+# 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507
+# 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368
+# 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353
+# 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055
+# 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294
+# 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491
+# 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324
+# 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204
+# 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936
+# 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731
+# 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594
+# 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609
+# 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534
+# 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847
+# 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802
+# 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385
+# 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206
+# 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654
+# 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027
+# 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953
+# 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654
+# 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132
+# 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266
+# 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893
+# 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042
+# 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446
+# 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371
+# 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535
+# 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535
+# 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416
+# 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983
+# 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327
+# 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491
+# 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163
+# 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179
+# 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701
+# 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254
+# 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671
+# 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209
+# 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134
+# 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895
+# 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373
+# 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821
+# 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403
+# 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254
+# 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627
+# 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658
+# 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971
+# 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628
+# 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404
+# 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494
+# 11:38:37 --- INFO: Training process end.
+# 11:38:37 --- INFO: Evaluating process start.
+# 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543}
+# 11:38:37 --- INFO: Evaluating process end.
+# 11:38:37 --- INFO: Predicting process start.
+# 11:38:37 --- INFO: Generating prediction results for mouse2_slice229.
+# 11:38:38 --- INFO: Generating prediction results for mouse2_slice300.
+# 11:38:38 --- INFO: Predicting process end.
+# 11:38:38 --- INFO: --------------------- GNN end ----------------------
+# 11:38:38 --- INFO: ----------------- Niche trajectory ------------------
+# 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array...
+# 11:38:38 --- INFO: Loading dataset.
+# 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100.
+# Processing...
+# 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# Done!
+# 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229
+# 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300
+# 11:38:39 --- INFO: Calculating NTScore for each niche.
+# 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force.
+# 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path.
+# 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level.
+# 11:38:39 --- INFO: Output NTScore tables.
+# 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization
16.1.3.1 load ONTraC results
- +The NTScore and binarized niche cluster info were stored in cell metadata
- -# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
-# <char> <char> <char> <char> <char> <char> <char> <int> <num>
-# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
-# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
-# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
-# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
-# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
-# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477# cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore
+# <char> <char> <char> <char> <char> <char> <char> <int> <num>
+# 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957
+# 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027
+# 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597
+# 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417
+# 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255
+# 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here.
- - + +16.1.3.2 niche cluster prob
-spatFeatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'niche cluster',
- expression_values = "prob",
- group_by = 'list_ID',
- feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
- point_border_col = 'gray'
- )spatFeatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'niche cluster',
+ expression_values = "prob",
+ group_by = 'list_ID',
+ feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob),
+ point_border_col = 'gray'
+ )
16.1.3.3 binarized niche cluster
-spatPlot2D(giotto_obj,
- spat_unit = "cell",
- group_by = 'list_ID',
- cell_color = "NicheCluster",
- color_as_factor = T,
- point_size = 1,
- point_border_stroke = NA,
- title="Niche cluster")spatPlot2D(giotto_obj,
+ spat_unit = "cell",
+ group_by = 'list_ID',
+ cell_color = "NicheCluster",
+ color_as_factor = T,
+ point_size = 1,
+ point_border_stroke = NA,
+ title="Niche cluster")
16.1.3.5 NT score
-spatPlot2D(gobject = giotto_obj,
- spat_unit = 'cell',
- feat_type = 'rna',
- group_by = 'list_ID',
- cell_color = "NTScore",
- color_as_factor = FALSE,
- cell_color_gradient = "turbo",
- point_size = 1,
- point_border_stroke = NA
- )spatPlot2D(gobject = giotto_obj,
+ spat_unit = 'cell',
+ feat_type = 'rna',
+ group_by = 'list_ID',
+ cell_color = "NTScore",
+ color_as_factor = FALSE,
+ cell_color_gradient = "turbo",
+ point_size = 1,
+ point_border_stroke = NA
+ )
15 Interoperability with other fr
15.1 Load Giotto object
To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process:
-
+
Load a Giotto mini Visium object, which will be used for demonstrating interoperability.
-
+
15.2 Seurat
@@ -533,99 +533,99 @@ 15.2 Seurat
15.2.1 Conversion of Giotto Object to Seurat Object
To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function
-
+
Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data.
-gToS <- giottoToSeuratV5(gobject = gobject,
- spat_unit = "cell")
-
-plot1 <- VlnPlot(gToS,
- features = "nCount_rna",
- pt.size = 0.1) +
- NoLegend()
-
-plot2 <- SpatialFeaturePlot(gToS,
- features = "nCount_rna",
- pt.size.factor = 2) +
- theme(legend.position = "right")
-
-wrap_plots(plot1, plot2)
+gToS <- giottoToSeuratV5(gobject = gobject,
+ spat_unit = "cell")
+
+plot1 <- VlnPlot(gToS,
+ features = "nCount_rna",
+ pt.size = 0.1) +
+ NoLegend()
+
+plot2 <- SpatialFeaturePlot(gToS,
+ features = "nCount_rna",
+ pt.size.factor = 2) +
+ theme(legend.position = "right")
+
+wrap_plots(plot1, plot2)
15.2.1.1 Apply SCTransform
We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function.
-
+
15.2.1.2 Dimension Reduction
We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object:
-
+
15.2.2 Conversion of Seurat object Back to Giotto Object
To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks.
-giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
- spatial_assay = "rna",
- dim_reduction = c("pca", "umap"),
- sp_network = "Delaunay_network",
- nn_network = c("sNN.pca", "custom_NN" ))
+giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
+ spatial_assay = "rna",
+ dim_reduction = c("pca", "umap"),
+ sp_network = "Delaunay_network",
+ nn_network = c("sNN.pca", "custom_NN" ))
15.2.2.1 Clustering and Plotting UMAP
Here we perform K-means clustering on the UMAP results obtained from the Seurat object:
-## k-means clustering
-giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
- dim_reduction_to_use = "pca")
-
-#Plot UMAP post-clustering to visualize kmeans
-graph2 <- Giotto::plotUMAP(
- gobject = giottoFromSeurat,
- cell_color = "kmeans",
- show_NN_network = TRUE,
- point_size = 2.5
-)
+## k-means clustering
+giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
+ dim_reduction_to_use = "pca")
+
+#Plot UMAP post-clustering to visualize kmeans
+graph2 <- Giotto::plotUMAP(
+ gobject = giottoFromSeurat,
+ cell_color = "kmeans",
+ show_NN_network = TRUE,
+ point_size = 2.5
+)
15.2.2.2 Spatial CoExpression
We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function:
-ranktest <- binSpect(giottoFromSeurat,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "Delaunay_network")
-
-ext_spatial_genes <- ranktest[1:300,]$feats
-
-spat_cor_netw_DT <- detectSpatialCorFeats(
- giottoFromSeurat,
- method = "network",
- spatial_network_name = "Delaunay_network",
- subset_feats = ext_spatial_genes)
-
-top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
- feats = "Dsp",
- show_top_feats = 10)
-
-spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
- name = "spat_netw_clus",
- k = 7)
-heatmSpatialCorFeats(
- giottoFromSeurat,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL),
- save_plot = TRUE,
- show_plot = TRUE,
- return_plot = FALSE,
- save_param = list(base_height = 6, base_width = 8, units = 'cm'))
+ranktest <- binSpect(giottoFromSeurat,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "Delaunay_network")
+
+ext_spatial_genes <- ranktest[1:300,]$feats
+
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ giottoFromSeurat,
+ method = "network",
+ spatial_network_name = "Delaunay_network",
+ subset_feats = ext_spatial_genes)
+
+top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
+ feats = "Dsp",
+ show_top_feats = 10)
+
+spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
+ name = "spat_netw_clus",
+ k = 7)
+heatmSpatialCorFeats(
+ giottoFromSeurat,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ return_plot = FALSE,
+ save_param = list(base_height = 6, base_width = 8, units = 'cm'))
@@ -635,60 +635,60 @@ 15.3 SpatialExperiment
15.1 Load Giotto object
To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process:
- +Load a Giotto mini Visium object, which will be used for demonstrating interoperability.
- +15.2 Seurat
@@ -533,99 +533,99 @@15.2 Seurat
15.2.1 Conversion of Giotto Object to Seurat Object
To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function
-
+
Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data.
-gToS <- giottoToSeuratV5(gobject = gobject,
- spat_unit = "cell")
-
-plot1 <- VlnPlot(gToS,
- features = "nCount_rna",
- pt.size = 0.1) +
- NoLegend()
-
-plot2 <- SpatialFeaturePlot(gToS,
- features = "nCount_rna",
- pt.size.factor = 2) +
- theme(legend.position = "right")
-
-wrap_plots(plot1, plot2)
+gToS <- giottoToSeuratV5(gobject = gobject,
+ spat_unit = "cell")
+
+plot1 <- VlnPlot(gToS,
+ features = "nCount_rna",
+ pt.size = 0.1) +
+ NoLegend()
+
+plot2 <- SpatialFeaturePlot(gToS,
+ features = "nCount_rna",
+ pt.size.factor = 2) +
+ theme(legend.position = "right")
+
+wrap_plots(plot1, plot2)
15.2.1.1 Apply SCTransform
We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function.
-
+
15.2.1.2 Dimension Reduction
We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object:
-
+
giottoToSeuratV5() functiongToS <- giottoToSeuratV5(gobject = gobject,
- spat_unit = "cell")
-
-plot1 <- VlnPlot(gToS,
- features = "nCount_rna",
- pt.size = 0.1) +
- NoLegend()
-
-plot2 <- SpatialFeaturePlot(gToS,
- features = "nCount_rna",
- pt.size.factor = 2) +
- theme(legend.position = "right")
-
-wrap_plots(plot1, plot2)gToS <- giottoToSeuratV5(gobject = gobject,
+ spat_unit = "cell")
+
+plot1 <- VlnPlot(gToS,
+ features = "nCount_rna",
+ pt.size = 0.1) +
+ NoLegend()
+
+plot2 <- SpatialFeaturePlot(gToS,
+ features = "nCount_rna",
+ pt.size.factor = 2) +
+ theme(legend.position = "right")
+
+wrap_plots(plot1, plot2)15.2.1.1 Apply SCTransform
We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function.
15.2.1.2 Dimension Reduction
We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object:
- +15.2.2 Conversion of Seurat object Back to Giotto Object
To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks.
giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
- spatial_assay = "rna",
- dim_reduction = c("pca", "umap"),
- sp_network = "Delaunay_network",
- nn_network = c("sNN.pca", "custom_NN" ))giottoFromSeurat <- seuratToGiottoV5(sobject = gToS,
+ spatial_assay = "rna",
+ dim_reduction = c("pca", "umap"),
+ sp_network = "Delaunay_network",
+ nn_network = c("sNN.pca", "custom_NN" ))15.2.2.1 Clustering and Plotting UMAP
Here we perform K-means clustering on the UMAP results obtained from the Seurat object:
-## k-means clustering
-giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
- dim_reduction_to_use = "pca")
-
-#Plot UMAP post-clustering to visualize kmeans
-graph2 <- Giotto::plotUMAP(
- gobject = giottoFromSeurat,
- cell_color = "kmeans",
- show_NN_network = TRUE,
- point_size = 2.5
-)## k-means clustering
+giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat,
+ dim_reduction_to_use = "pca")
+
+#Plot UMAP post-clustering to visualize kmeans
+graph2 <- Giotto::plotUMAP(
+ gobject = giottoFromSeurat,
+ cell_color = "kmeans",
+ show_NN_network = TRUE,
+ point_size = 2.5
+)
15.2.2.2 Spatial CoExpression
We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function:
ranktest <- binSpect(giottoFromSeurat,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "Delaunay_network")
-
-ext_spatial_genes <- ranktest[1:300,]$feats
-
-spat_cor_netw_DT <- detectSpatialCorFeats(
- giottoFromSeurat,
- method = "network",
- spatial_network_name = "Delaunay_network",
- subset_feats = ext_spatial_genes)
-
-top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
- feats = "Dsp",
- show_top_feats = 10)
-
-spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
- name = "spat_netw_clus",
- k = 7)
-heatmSpatialCorFeats(
- giottoFromSeurat,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL),
- save_plot = TRUE,
- show_plot = TRUE,
- return_plot = FALSE,
- save_param = list(base_height = 6, base_width = 8, units = 'cm'))ranktest <- binSpect(giottoFromSeurat,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "Delaunay_network")
+
+ext_spatial_genes <- ranktest[1:300,]$feats
+
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ giottoFromSeurat,
+ method = "network",
+ spatial_network_name = "Delaunay_network",
+ subset_feats = ext_spatial_genes)
+
+top10_genes <- showSpatialCorFeats(spat_cor_netw_DT,
+ feats = "Dsp",
+ show_top_feats = 10)
+
+spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT,
+ name = "spat_netw_clus",
+ k = 7)
+heatmSpatialCorFeats(
+ giottoFromSeurat,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ return_plot = FALSE,
+ save_param = list(base_height = 6, base_width = 8, units = 'cm'))
To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries.
-library(SpatialExperiment)
-library(ggspavis)
-library(pheatmap)
-library(scater)
-library(scran)
-library(nnSVG)library(SpatialExperiment)
+library(ggspavis)
+library(pheatmap)
+library(scater)
+library(scran)
+library(nnSVG)15.3.1 Convert Giotto Object to SpatialExperiment Object
To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function.
The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use:
- +15.3.1.1 Identify top spatially variable genes with nnSVG
We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate:
-# One of the assays should be "logcounts"
-# We rename the normalized assay to "logcounts"
-assayNames(spe)[[2]] <- "logcounts"
-
-# Create model matrix for leiden clustering labels
-X <- model.matrix(~ colData(spe)$leiden_clus)
-dim(X)# One of the assays should be "logcounts"
+# We rename the normalized assay to "logcounts"
+assayNames(spe)[[2]] <- "logcounts"
+
+# Create model matrix for leiden clustering labels
+X <- model.matrix(~ colData(spe)$leiden_clus)
+dim(X)- Run nnSVG
This step will take several minutes to run
- +15.3.2 Conversion of SpatialExperiment object back to Giotto
We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object.
giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-
-giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
- python_path = NULL,
- sp_network = "Delaunay_network")
-print(giottoFromSPE)giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+
+giottoFromSPE <- spatialExperimentToGiotto(spe = spe,
+ python_path = NULL,
+ sp_network = "Delaunay_network")
+print(giottoFromSPE)15.3.2.1 Plotting top genes from nnSVG in Giotto
Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object.
-ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
- output = "data.table")
-
-ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_IDspatFeatPlot2D(giottoFromSPE,
- expression_values = "scaled_rna_cell",
- feats = ext_spatial_genes[1:4],
- point_size = 2)ext_spatial_genes <- getFeatureMetadata(giottoFromSPE,
+ output = "data.table")
+
+ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_IDspatFeatPlot2D(giottoFromSPE,
+ expression_values = "scaled_rna_cell",
+ feats = ext_spatial_genes[1:4],
+ point_size = 2)
15.4 AnnData
15.4.1 Load Required Libraries
To start, we need to load the necessary libraries, including reticulate for interfacing with Python.
-
+
reticulate for interfacing with Python.15.4.2 Specify Path for Results
First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions.
-# Specify path to which results may be saved
-results_directory <- "results/03_session4/giotto_anndata_conversion/"
-
-instrs <- showGiottoInstructions(gobject)
-
-mini_gobject <- replaceGiottoInstructions(gobject = gobject,
- instructions = instrs)# Specify path to which results may be saved
+results_directory <- "results/03_session4/giotto_anndata_conversion/"
+
+instrs <- showGiottoInstructions(gobject)
+
+mini_gobject <- replaceGiottoInstructions(gobject = gobject,
+ instructions = instrs)15.4.2.1 Create Default kNN Network
We will create a k-nearest neighbor (kNN) network using mostly default parameters.
- +15.4.3 Giotto To AnnData
To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData()
Next, we import scanpy and perform a series of preprocessing steps on the AnnData object.
-scanpy <- import("scanpy")
-
-adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
-
-# Normalize total counts per cell
-scanpy$pp$normalize_total(adata, target_sum=1e4)
-
-# Log-transform the data
-scanpy$pp$log1p(adata)
-
-# Perform PCA
-scanpy$pp$pca(adata, n_comps=40L)
-
-# Compute the neighborhood graph
-scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
-
-# Run UMAP
-scanpy$tl$umap(adata)
-
-# Save the processed AnnData object
-adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
-
-processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"scanpy <- import("scanpy")
+
+adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad")
+
+# Normalize total counts per cell
+scanpy$pp$normalize_total(adata, target_sum=1e4)
+
+# Log-transform the data
+scanpy$pp$log1p(adata)
+
+# Perform PCA
+scanpy$pp$pca(adata, n_comps=40L)
+
+# Compute the neighborhood graph
+scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L)
+
+# Run UMAP
+scanpy$tl$umap(adata)
+
+# Save the processed AnnData object
+adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad")
+
+processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad"15.4.4 Convert AnnData to Giotto
Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto.
- +15.4.4.1 UMAP Visualization
Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object.
Giotto::plotUMAP(
- gobject = giottoFromAnndata,
- dim_reduction_name = "umap.ad",
- cell_color = "leiden_clus",
- point_size = 3
-)Giotto::plotUMAP(
+ gobject = giottoFromAnndata,
+ dim_reduction_name = "umap.ad",
+ cell_color = "leiden_clus",
+ point_size = 3
+)
15.5 Create mini Vizgen object
-mini_gobject <- loadGiottoMini(dataset = "vizgen",
- python_path = NULL)
-
-mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
- instructions = instrs)mini_gobject <- createNearestNetwork(gobject = mini_gobject,
- spat_unit = "aggregate",
- feat_type = "rna",
- type = "kNN",
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap",
- k = 6,
- name = "new_network")mini_gobject <- loadGiottoMini(dataset = "vizgen",
+ python_path = NULL)
+
+mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject,
+ instructions = instrs)mini_gobject <- createNearestNetwork(gobject = mini_gobject,
+ spat_unit = "aggregate",
+ feat_type = "rna",
+ type = "kNN",
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap",
+ k = 6,
+ name = "new_network")Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details.
4.3.1 Giotto installation
4.3.1.1 System prerequisites
+-
+
- for windows, Rtools needs to be installed +
- a major dependency terra needs
GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3on linux
+
4.3.1.2 Installation of released version
+To install the currently released version of Giotto in a single step:
+This should automatically install all the Giotto dependencies and other Giotto module packages.
+4.3.1.3 Installation of dev branch Giotto packages
+You can install dev branch versions by using devtools::install_github()
Core module dev branchs:
+-
+
- “drieslab/Giotto@suite_dev” +
- “drieslab/GiottoVisuals@dev” +
- “drieslab/GiottoClass@dev” +
- “drieslab/GiottoUtils@dev” +
pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages.
+4.3.1.4 Common install issues
+If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work.
+ +If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it.
## # Downloading packages -------------------------------------------------------
+## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s]
+## Successfully downloaded 1 package in 1.2 seconds.
+##
+## The following package(s) will be installed:
+## - irlba [2.3.5.1]
+## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu".
+##
+## # Installing packages --------------------------------------------------------
+## - Installing irlba ... OK [built from source and cached in 5.7s]
+## Successfully installed 1 package in 5.7 seconds.4.3.2 Python environment
4.3.2.1 Default installation
In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment.
- +Two things are needed for python to work:
- A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. @@ -580,17 +619,17 @@
4.3.2.1 Default installation
4.3.2.2 Custom installs
Custom python environments can be made by first setting up a new environment and establishing the name and python version to use.
-
+
Following that, one or more python packages to install can be added to the environment.
-reticulate::py_install(
- pip = TRUE,
- envname = '[name of env]',
- packages = c(
- "package1",
- "package2",
- "..."
- )
-)
+reticulate::py_install(
+ pip = TRUE,
+ envname = '[name of env]',
+ packages = c(
+ "package1",
+ "package2",
+ "..."
+ )
+)
Once an environment has been set up, Giotto can hook into it.
reticulate::py_install(
- pip = TRUE,
- envname = '[name of env]',
- packages = c(
- "package1",
- "package2",
- "..."
- )
-)reticulate::py_install(
+ pip = TRUE,
+ envname = '[name of env]',
+ packages = c(
+ "package1",
+ "package2",
+ "..."
+ )
+)
@@ -612,17 +651,56 @@
4.3.2.3 Using a specific environm
Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto.
You would run file.edit("~/.Rprofile") and then add options("giotto.py_path" = "[path to env or envname]") as a line so that it is automatically set at the start of each session.
If a specific environment should only be used a couple times then method 1 is easiest:
- +To check which conda environments exist on your machine:
- +Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running:
- +4.4 Giotto instructions
-Giotto
+Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are:
-
+
- python_path - when set, will activate a python environment +
- save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render. +
- save_plot - whether to save plots to the
save_dir
+ - return_plot - whether to return the plot objects. When FALSE, only NULL is returned +
- show_plot - whether to show the plot in the viewer +
These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function.
library(Giotto)
+save_dir <- "results/01_session2/"
+
+# this call will also intialize the python env
+instrs <- createGiottoInstructions(
+ save_dir = save_dir, # working directory is the default
+ show_plot = FALSE,
+ save_plot = TRUE,
+ return_plot = FALSE,
+ python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default
+)
+force(instrs)Giotto object creation functions all have an instructions param for passing in instructions objects.
+giotto objects will also respond to the instructions() generic.
test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object
+
+# example plot
+g <- GiottoData::loadGiottoMini("visium")
+instructions(g) <- instrs
+instructions(g, "show_plot") # instructions say not to plot to viewer
+spatPlot2D(g, show_image = TRUE, image_name = "image")
+# instead it will directly write to the results folderAs an example, you can also set individual instructions
+ +
+
+
++Figure 4.2: example image output +
+14 Multi-omics integration
14.1 The CytAssist technology
The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image.
-
+
Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics.
The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument.
-
+
Figure 14.2: CytAssist workflow. Source: 10X genomics.
@@ -542,19 +542,19 @@
14.2 Introduction to the spatial
14.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/03_session3")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
+dir.create("data/03_session3")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”.
-
+
14.4 Create the Giotto object
@@ -564,124 +564,124 @@ 14.4 Create the Giotto objectx,y(,z) coordinates for cells or spots (or the path to)
createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/03_session3/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-# Provide the path to the data folder
-data_path <- "data/03_session3/"
-
-# Create object directly from the data folder
-visium_tonsil <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/03_session3/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+# Provide the path to the data folder
+data_path <- "data/03_session3/"
+
+# Create object directly from the data folder
+visium_tonsil <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
- Print the information of the object, note that both rna and protein are listed in the expression slot.
-
+
14.5 Subset on spots that were covered by tissue
-spatPlot2D(
- gobject = visium_tonsil,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE,
- image_name = "image"
-)
-
+spatPlot2D(
+ gobject = visium_tonsil,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE,
+ image_name = "image"
+)
+
Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object.
-
+
14.6 RNA processing
- Run the Filtering, normalization, and statistics steps using only the RNA feature.
-visium_tonsil <- filterGiotto(
- gobject = visium_tonsil,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil)
+visium_tonsil <- filterGiotto(
+ gobject = visium_tonsil,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1000,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil)
- Dimension reduction
Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs.
-visium_tonsil <- calculateHVF(gobject = visium_tonsil)
-
-visium_tonsil <- runPCA(gobject = visium_tonsil)
+visium_tonsil <- calculateHVF(gobject = visium_tonsil)
+
+visium_tonsil <- runPCA(gobject = visium_tonsil)
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the RNA-based Leiden clusters.
-
+
- Visualization
Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot.
-plotUMAP(gobject = visium_tonsil,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters.
Plot the spatial distribution of the RNA-based Leiden cluster per spot.
-spatPlot2D(gobject = visium_tonsil,
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.5: Spatial distribution of RNA-based Leiden clusters.
@@ -693,80 +693,80 @@
14.7 Protein processingvisium_tonsil <- filterGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein")
+visium_tonsil <- filterGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein")
- Dimension reduction
Calculate the principal components using all the proteins available in the dataset.
-
+
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the Protein-based Leiden clusters.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- resolution = 1,
- n_iterations = 1000)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ resolution = 1,
+ n_iterations = 1000)
- Visualization
Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters.
Plot the spatial distribution of the Protein-based Leiden clusters.
-spatPlot2D(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.7: Spatial distribution of Protein-based Leiden clusters.
@@ -779,68 +779,68 @@
14.8 Multi-omics integrationCalculate kNN
Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type.
-## RNA modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
-
-## Protein modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
+## RNA modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
+
+## Protein modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
- Run WNN
Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object.
-visium_tonsil <- runWNN(visium_tonsil,
- spat_unit = "cell",
- modality_1 = "rna",
- modality_2 = "protein",
- pca_name_modality_1 = "pca",
- pca_name_modality_2 = "protein.pca",
- k = 20,
- integrated_feat_type = NULL,
- matrix_result_name = NULL,
- w_name_modality_1 = NULL,
- w_name_modality_2 = NULL,
- verbose = TRUE)
+visium_tonsil <- runWNN(visium_tonsil,
+ spat_unit = "cell",
+ modality_1 = "rna",
+ modality_2 = "protein",
+ pca_name_modality_1 = "pca",
+ pca_name_modality_2 = "protein.pca",
+ k = 20,
+ integrated_feat_type = NULL,
+ matrix_result_name = NULL,
+ w_name_modality_1 = NULL,
+ w_name_modality_2 = NULL,
+ verbose = TRUE)
- Run Integrated umap
Calculate the UMAP using the weights of each feature per spot.
-visium_tonsil <- runIntegratedUMAP(visium_tonsil,
- modality1 = "rna",
- modality2 = "protein",
- spread = 7,
- min_dist = 1,
- force = FALSE)
+visium_tonsil <- runIntegratedUMAP(visium_tonsil,
+ modality1 = "rna",
+ modality2 = "protein",
+ spread = 7,
+ min_dist = 1,
+ force = FALSE)
- Calculate integrated clusters
Calculate the multiomics-based Leiden clusters using the weights of each feature per spot.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- nn_network_to_use = "kNN",
- network_name = "integrated_kNN",
- name = "integrated_leiden_clus",
- resolution = 1)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ nn_network_to_use = "kNN",
+ network_name = "integrated_kNN",
+ name = "integrated_leiden_clus",
+ resolution = 1)
- Visualize the integrated UMAP
Plot the integrated UMAP and color the spots using the integrated Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- dim_reduction_name = "integrated.umap",
- point_size = 2,
- title = "Integrated UMAP using Integrated Leiden clusters")
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ dim_reduction_name = "integrated.umap",
+ point_size = 2,
+ title = "Integrated UMAP using Integrated Leiden clusters")
+
Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters.
@@ -850,14 +850,14 @@
14.8 Multi-omics integrationVisualize spatial plot with integrated clusters
Plot the spatial distribution of the integrated Leiden clusters.
-spatPlot2D(visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- point_size = 3,
- show_image = TRUE,
- title = "Integrated Leiden clustering")
-
+spatPlot2D(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ point_size = 3,
+ show_image = TRUE,
+ title = "Integrated Leiden clustering")
+
Figure 14.9: Spatial distribution of the integrated Leiden clusters.
@@ -866,85 +866,85 @@
14.8 Multi-omics integration
14.9 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
- [7] matrixStats_1.3.0 compiler_4.4.1
- [9] png_0.1-8 systemfonts_1.1.0
- [11] vctrs_0.6.5 reshape2_1.4.4
- [13] stringr_1.5.1 pkgconfig_2.0.3
- [15] SpatialExperiment_1.14.0 crayon_1.5.3
- [17] fastmap_1.2.0 backports_1.5.0
- [19] magick_2.8.4 XVector_0.44.0
- [21] labeling_0.4.3 utf8_1.2.4
- [23] rmarkdown_2.27 UCSC.utils_1.0.0
- [25] ragg_1.3.2 purrr_1.0.2
- [27] xfun_0.46 beachmat_2.20.0
- [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
- [31] jsonlite_1.8.8 DelayedArray_0.30.1
- [33] BiocParallel_1.38.0 terra_1.7-78
- [35] irlba_2.3.5.1 parallel_4.4.1
- [37] R6_2.5.1 stringi_1.8.4
- [39] RColorBrewer_1.1-3 reticulate_1.38.0
- [41] parallelly_1.37.1 GenomicRanges_1.56.1
- [43] scattermore_1.2 Rcpp_1.0.13
- [45] bookdown_0.40 SummarizedExperiment_1.34.0
- [47] knitr_1.48 future.apply_1.11.2
- [49] R.utils_2.12.3 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
+ [7] matrixStats_1.3.0 compiler_4.4.1
+ [9] png_0.1-8 systemfonts_1.1.0
+ [11] vctrs_0.6.5 reshape2_1.4.4
+ [13] stringr_1.5.1 pkgconfig_2.0.3
+ [15] SpatialExperiment_1.14.0 crayon_1.5.3
+ [17] fastmap_1.2.0 backports_1.5.0
+ [19] magick_2.8.4 XVector_0.44.0
+ [21] labeling_0.4.3 utf8_1.2.4
+ [23] rmarkdown_2.27 UCSC.utils_1.0.0
+ [25] ragg_1.3.2 purrr_1.0.2
+ [27] xfun_0.46 beachmat_2.20.0
+ [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
+ [31] jsonlite_1.8.8 DelayedArray_0.30.1
+ [33] BiocParallel_1.38.0 terra_1.7-78
+ [35] irlba_2.3.5.1 parallel_4.4.1
+ [37] R6_2.5.1 stringi_1.8.4
+ [39] RColorBrewer_1.1-3 reticulate_1.38.0
+ [41] parallelly_1.37.1 GenomicRanges_1.56.1
+ [43] scattermore_1.2 Rcpp_1.0.13
+ [45] bookdown_0.40 SummarizedExperiment_1.34.0
+ [47] knitr_1.48 future.apply_1.11.2
+ [49] R.utils_2.12.3 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/reference-keys.txt b/reference-keys.txt
index a17604d..555f956 100644
--- a/reference-keys.txt
+++ b/reference-keys.txt
@@ -4,115 +4,116 @@ fig:unnamed-chunk-14
fig:unnamed-chunk-15
fig:unnamed-chunk-16
fig:unnamed-chunk-18
-fig:unnamed-chunk-30
-fig:unnamed-chunk-35
-fig:unnamed-chunk-38
-fig:unnamed-chunk-40
+fig:unnamed-chunk-32
+fig:unnamed-chunk-37
fig:unnamed-chunk-42
-fig:unnamed-chunk-44
+fig:unnamed-chunk-45
+fig:unnamed-chunk-47
fig:unnamed-chunk-49
fig:unnamed-chunk-51
-fig:unnamed-chunk-53
-fig:unnamed-chunk-55
-fig:unnamed-chunk-59
-fig:unnamed-chunk-61
-fig:unnamed-chunk-63
+fig:unnamed-chunk-56
+fig:unnamed-chunk-58
+fig:unnamed-chunk-60
+fig:unnamed-chunk-62
fig:unnamed-chunk-66
-fig:unnamed-chunk-69
-fig:unnamed-chunk-74
+fig:unnamed-chunk-68
+fig:unnamed-chunk-70
+fig:unnamed-chunk-73
fig:unnamed-chunk-76
-fig:unnamed-chunk-78
-fig:unnamed-chunk-80
-fig:unnamed-chunk-82
-fig:unnamed-chunk-84
+fig:unnamed-chunk-81
+fig:unnamed-chunk-83
+fig:unnamed-chunk-85
fig:unnamed-chunk-87
+fig:unnamed-chunk-89
+fig:unnamed-chunk-91
fig:unnamed-chunk-94
-fig:unnamed-chunk-96
-fig:unnamed-chunk-98
fig:unnamed-chunk-101
fig:unnamed-chunk-103
fig:unnamed-chunk-105
+fig:unnamed-chunk-108
+fig:unnamed-chunk-110
fig:unnamed-chunk-112
-fig:unnamed-chunk-114
-fig:unnamed-chunk-118
-fig:unnamed-chunk-120
-fig:unnamed-chunk-123
-fig:unnamed-chunk-128
-fig:unnamed-chunk-131
-fig:unnamed-chunk-134
-fig:unnamed-chunk-137
+fig:unnamed-chunk-119
+fig:unnamed-chunk-121
+fig:unnamed-chunk-125
+fig:unnamed-chunk-127
+fig:unnamed-chunk-130
+fig:unnamed-chunk-135
+fig:unnamed-chunk-138
fig:unnamed-chunk-141
-fig:unnamed-chunk-143
-fig:unnamed-chunk-147
+fig:unnamed-chunk-144
+fig:unnamed-chunk-148
fig:unnamed-chunk-150
-fig:unnamed-chunk-152
+fig:unnamed-chunk-154
+fig:unnamed-chunk-157
fig:unnamed-chunk-159
-fig:unnamed-chunk-161
-fig:unnamed-chunk-163
-fig:unnamed-chunk-165
-fig:unnamed-chunk-167
-fig:unnamed-chunk-169
-fig:unnamed-chunk-171
+fig:unnamed-chunk-166
+fig:unnamed-chunk-168
+fig:unnamed-chunk-170
+fig:unnamed-chunk-172
fig:unnamed-chunk-174
-fig:unnamed-chunk-175
fig:unnamed-chunk-176
-fig:unnamed-chunk-185
-fig:unnamed-chunk-191
-fig:unnamed-chunk-194
+fig:unnamed-chunk-178
+fig:unnamed-chunk-181
+fig:unnamed-chunk-182
+fig:unnamed-chunk-183
+fig:unnamed-chunk-192
+fig:unnamed-chunk-198
fig:unnamed-chunk-201
-fig:unnamed-chunk-204
-fig:unnamed-chunk-206
-fig:unnamed-chunk-210
-fig:unnamed-chunk-212
-fig:unnamed-chunk-215
+fig:unnamed-chunk-208
+fig:unnamed-chunk-211
+fig:unnamed-chunk-213
fig:unnamed-chunk-217
fig:unnamed-chunk-219
-fig:unnamed-chunk-221
+fig:unnamed-chunk-222
fig:unnamed-chunk-224
-fig:unnamed-chunk-225
-fig:unnamed-chunk-227
-fig:unnamed-chunk-229
-fig:unnamed-chunk-230
+fig:unnamed-chunk-226
+fig:unnamed-chunk-228
+fig:unnamed-chunk-231
fig:unnamed-chunk-232
fig:unnamed-chunk-234
fig:unnamed-chunk-236
-fig:unnamed-chunk-275
-fig:unnamed-chunk-279
-fig:unnamed-chunk-281
-fig:unnamed-chunk-283
+fig:unnamed-chunk-237
+fig:unnamed-chunk-239
+fig:unnamed-chunk-241
+fig:unnamed-chunk-243
+fig:unnamed-chunk-282
fig:unnamed-chunk-286
fig:unnamed-chunk-288
fig:unnamed-chunk-290
-fig:unnamed-chunk-292
-fig:unnamed-chunk-294
+fig:unnamed-chunk-293
+fig:unnamed-chunk-295
fig:unnamed-chunk-297
fig:unnamed-chunk-299
fig:unnamed-chunk-301
-fig:unnamed-chunk-303
-fig:unnamed-chunk-305
-fig:unnamed-chunk-365
-fig:unnamed-chunk-366
+fig:unnamed-chunk-304
+fig:unnamed-chunk-306
+fig:unnamed-chunk-308
+fig:unnamed-chunk-310
+fig:unnamed-chunk-312
fig:unnamed-chunk-372
+fig:unnamed-chunk-373
fig:unnamed-chunk-379
-fig:unnamed-chunk-381
-fig:unnamed-chunk-387
-fig:unnamed-chunk-389
-fig:unnamed-chunk-395
-fig:unnamed-chunk-397
-fig:unnamed-chunk-462
-fig:unnamed-chunk-465
+fig:unnamed-chunk-386
+fig:unnamed-chunk-388
+fig:unnamed-chunk-394
+fig:unnamed-chunk-396
+fig:unnamed-chunk-402
+fig:unnamed-chunk-404
fig:unnamed-chunk-469
-fig:unnamed-chunk-470
fig:unnamed-chunk-472
-fig:unnamed-chunk-474
fig:unnamed-chunk-476
-fig:unnamed-chunk-478
+fig:unnamed-chunk-477
fig:unnamed-chunk-479
fig:unnamed-chunk-481
+fig:unnamed-chunk-483
fig:unnamed-chunk-485
+fig:unnamed-chunk-486
fig:unnamed-chunk-488
-fig:unnamed-chunk-490
fig:unnamed-chunk-492
+fig:unnamed-chunk-495
+fig:unnamed-chunk-497
+fig:unnamed-chunk-499
giotto-suite-workshop-2024
instructors
topics-and-schedule
@@ -153,6 +154,10 @@ presentation-1
ecosystem
installation-python-environment
giotto-installation-1
+system-prerequisites
+installation-of-released-version
+installation-of-dev-branch-giotto-packages
+common-install-issues
python-environment
default-installation
custom-installs
diff --git a/search_index.json b/search_index.json
index 2736898..72ab7c3 100644
--- a/search_index.json
+++ b/search_index.json
@@ -1 +1 @@
-[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: Ensure your system To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak. pak::pak("drieslab/Giotto") 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
+[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: 4.3.1.1 System prerequisites for windows, Rtools needs to be installed a major dependency terra needs GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux 4.3.1.2 Installation of released version To install the currently released version of Giotto in a single step: pak::pak("drieslab/Giotto") This should automatically install all the Giotto dependencies and other Giotto module packages. 4.3.1.3 Installation of dev branch Giotto packages You can install dev branch versions by using devtools::install_github() Core module dev branchs: “drieslab/Giotto@suite_dev” “drieslab/GiottoVisuals@dev” “drieslab/GiottoClass@dev” “drieslab/GiottoUtils@dev” devtools::install_github("drieslab/GiottoClass@dev") pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages. 4.3.1.4 Common install issues If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work. devtools::install_version("Matrix", version = "1.6-5") If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it. install.packages("irlba", type = "source") ## # Downloading packages ------------------------------------------------------- ## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s] ## Successfully downloaded 1 package in 1.2 seconds. ## ## The following package(s) will be installed: ## - irlba [2.3.5.1] ## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu". ## ## # Installing packages -------------------------------------------------------- ## - Installing irlba ... OK [built from source and cached in 5.7s] ## Successfully installed 1 package in 5.7 seconds. 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are: python_path - when set, will activate a python environment save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render. save_plot - whether to save plots to the save_dir return_plot - whether to return the plot objects. When FALSE, only NULL is returned show_plot - whether to show the plot in the viewer These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function. library(Giotto) save_dir <- "results/01_session2/" # this call will also intialize the python env instrs <- createGiottoInstructions( save_dir = save_dir, # working directory is the default show_plot = FALSE, save_plot = TRUE, return_plot = FALSE, python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default ) force(instrs) Giotto object creation functions all have an instructions param for passing in instructions objects. giotto objects will also respond to the instructions() generic. test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object # example plot g <- GiottoData::loadGiottoMini("visium") instructions(g) <- instrs instructions(g, "show_plot") # instructions say not to plot to viewer spatPlot2D(g, show_image = TRUE, image_name = "image") # instead it will directly write to the results folder As an example, you can also set individual instructions instructions(g, "show_plot") <- TRUE spatPlot2D(g, show_image = TRUE, image_name = "image") Figure 4.2: example image output "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
diff --git a/spatial-multi-modal-analysis.html b/spatial-multi-modal-analysis.html
index 5bdab74..c78f952 100644
--- a/spatial-multi-modal-analysis.html
+++ b/spatial-multi-modal-analysis.html
@@ -563,48 +563,48 @@ 13.2.3 Spatial classes:# load in data
-library(Giotto)
-
-g <- GiottoData::loadGiottoMini("vizgen")
-
-activeSpatUnit(g) <- "aggregate"
-
-gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE)
-
-gimg <- getGiottoImage(g)
+
13.3 Examples of the simple transforms with a giottoPolygon
-rain <- rainbow(nrow(gpoly))
-
-line_width <- 0.3
-
-# par to setup the grid plotting layout
-p <- par(no.readonly = TRUE)
-par(mfrow=c(3,3))
-gpoly |>
- plot(main = "no transform", col = rain, lwd = line_width)
-
-gpoly |> spatShift(dx = 1000) |>
- plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
-
-gpoly |> spin(45) |>
- plot(main = "spin(45)", col = rain, lwd = line_width)
-
-gpoly |> rescale(fx = 10, fy = 6) |>
- plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
-
-gpoly |> flip(direction = "vertical") |>
- plot(main = "flip()", col = rain, lwd = line_width)
-
-gpoly |> t() |>
- plot(main = "t()", col = rain, lwd = line_width)
-
-gpoly |> shear(fx = 0.5) |>
- plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
-par(p)
+rain <- rainbow(nrow(gpoly))
+
+line_width <- 0.3
+
+# par to setup the grid plotting layout
+p <- par(no.readonly = TRUE)
+par(mfrow=c(3,3))
+gpoly |>
+ plot(main = "no transform", col = rain, lwd = line_width)
+
+gpoly |> spatShift(dx = 1000) |>
+ plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
+
+gpoly |> spin(45) |>
+ plot(main = "spin(45)", col = rain, lwd = line_width)
+
+gpoly |> rescale(fx = 10, fy = 6) |>
+ plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
+
+gpoly |> flip(direction = "vertical") |>
+ plot(main = "flip()", col = rain, lwd = line_width)
+
+gpoly |> t() |>
+ plot(main = "t()", col = rain, lwd = line_width)
+
+gpoly |> shear(fx = 0.5) |>
+ plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
+par(p)
@@ -617,20 +617,20 @@ 13.4 Affine transforms# create affine2d
-aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
+# create affine2d
+aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set.
-
-# append several simple transforms
-aff <- aff |>
- spatShift(dx = 1000) |>
- spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- flip(direction = "vertical") |>
- t() |>
- shear(fx = 0.5)
-force(aff)
+
+# append several simple transforms
+aff <- aff |>
+ spatShift(dx = 1000) |>
+ spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ flip(direction = "vertical") |>
+ t() |>
+ shear(fx = 0.5)
+force(aff)
<affine2d>
anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax)
rotate : -0.785398163397448 (rad)
@@ -639,35 +639,35 @@ 13.4 Affine transformsplot(aff)
+
We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests.
-
+
13.5 Image transforms
Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting.
spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image.
-# affine transform of image info matches with polygon info
-gimg |> affine(aff) |> plot()
-gpoly |> affine(aff) |>
- plot(add = TRUE,
- border = "cyan",
- lwd = 0.3)
+# affine transform of image info matches with polygon info
+gimg |> affine(aff) |> plot()
+gpoly |> affine(aff) |>
+ plot(add = TRUE,
+ border = "cyan",
+ lwd = 0.3)
-# affine of the giotto object
-g |> affine(aff) |>
- spatInSituPlotPoints(
- show_image = TRUE,
- feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
- feats_color_code = rainbow(4),
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- point_size = 0.1,
- use_overlap = FALSE
- )
+# affine of the giotto object
+g |> affine(aff) |>
+ spatInSituPlotPoints(
+ show_image = TRUE,
+ feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
+ feats_color_code = rainbow(4),
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ point_size = 0.1,
+ use_overlap = FALSE
+ )
Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used.
@@ -687,256 +687,256 @@ 13.6.1 Example dataset: Xenium Br
– cooresponding centroid locations
The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance.
-library(Giotto)
-instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
- save_plot = TRUE,
- show_plot = TRUE)
-options(timeout = 999999)
-download_dir <-file.path(getwd(),'/data/03_session2/')
-destfile <- file.path(download_dir,'Multimodal_registration.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
-Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
-Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
+library(Giotto)
+instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
+ save_plot = TRUE,
+ show_plot = TRUE)
+options(timeout = 999999)
+download_dir <-file.path(getwd(),'/data/03_session2/')
+destfile <- file.path(download_dir,'Multimodal_registration.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
+Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
+Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
13.6.2 Target Coordinate system
Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system.
-xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
-xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
-xen_cell_pl
+xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
+xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+xen_cell_pl
13.6.3 Visium to register
Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration
-G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
- gene_column_index = 2,
- png_name = 'tissue_hires_image.png',
- instructions = NULL)
-# In the meantime, calculate statistics for easier plot showing
-G_visium <- normalizeGiotto(G_visium)
-G_visium <- addStatistics(G_visium)
-V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
-V_origin
+G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
+ gene_column_index = 2,
+ png_name = 'tissue_hires_image.png',
+ instructions = NULL)
+# In the meantime, calculate statistics for easier plot showing
+G_visium <- normalizeGiotto(G_visium)
+G_visium <- addStatistics(G_visium)
+V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
+V_origin
The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform.
-# create affine2d
-aff <- affine(diag(c(1,1)))
-aff <- aff |>
- spin(90) |>
- flip(direction = "horizontal")
-force(aff)
-
-# Apply the transform
-V_tansformed <- affine(G_visium,aff)
-spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
-spatplot_to_register
+# create affine2d
+aff <- affine(diag(c(1,1)))
+aff <- aff |>
+ spin(90) |>
+ flip(direction = "horizontal")
+force(aff)
+
+# Apply the transform
+V_tansformed <- affine(G_visium,aff)
+spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
+spatplot_to_register
Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images.
Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app.
-
+
Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result.
-landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
-affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
-
-V_final <- affine(G_visium,affine_mtx %*% aff@affine)
-
-spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
-spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
+affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
+
+V_final <- affine(G_visium,affine_mtx %*% aff@affine)
+
+spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
+spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.3.1 Create Pseudo Visium dataset for comparison
Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance.
-V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
-spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
-spatPlot2D(V_final,
- show_network = T,
- network_color = 'blue',
- point_size = 1)
-center_to_center <- min(spat_network$distance)
-radius <- center_to_center*55/200
+V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
+spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
+spatPlot2D(V_final,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1)
+center_to_center <- min(spat_network$distance)
+radius <- center_to_center*55/200
Now we get the Pseudo Visium polygons
-Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
-stamp_dt <- circleVertices(radius = radius, npoints = 100)
-pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
-pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
-plot(pseudo_visium_poly)
+Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
+stamp_dt <- circleVertices(radius = radius, npoints = 100)
+pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
+pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
+plot(pseudo_visium_poly)
Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points
-xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
-gpoints <- createGiottoPoints(xen_transcripts)
-Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
- gpolygons = list('visium' = pseudo_visium_poly))
+xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
+gpoints <- createGiottoPoints(xen_transcripts)
+Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
+ gpolygons = list('visium' = pseudo_visium_poly))
Get gene expression information by overlapping polygon to points
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'visium')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "visium",
- feat_info = "rna",
- aggr_function = "sum")
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'visium')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "visium",
+ feat_info = "rna",
+ aggr_function = "sum")
Manipulate the expression for plotting
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'visium',
- name = 'plot')
-
-spatFeatPlot2D(Xen_obj,
- point_size = 3.5,
- expression_values = 'plot',
- show_image = F,
- feats = 'ERBB2')
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ name = 'plot')
+
+spatFeatPlot2D(Xen_obj,
+ point_size = 3.5,
+ expression_values = 'plot',
+ show_image = F,
+ feats = 'ERBB2')
Subset the registered Visium and plot same gene
-#get the extent of giotto points, xmin, xmax, ymin, ymax
-subset_extent <- ext(gpoints@spatVector)
-sub_visium <- subsetGiottoLocs(V_final,
- x_min = subset_extent[1],
- x_max = subset_extent[2],
- y_min = subset_extent[3],
- y_max = subset_extent[4])
-spatFeatPlot2D(sub_visium,
- point_size = 2,
- expression_values = 'scaled',
- show_image = F,
- feats = 'ERBB2')
+#get the extent of giotto points, xmin, xmax, ymin, ymax
+subset_extent <- ext(gpoints@spatVector)
+sub_visium <- subsetGiottoLocs(V_final,
+ x_min = subset_extent[1],
+ x_max = subset_extent[2],
+ y_min = subset_extent[3],
+ y_max = subset_extent[4])
+spatFeatPlot2D(sub_visium,
+ point_size = 2,
+ expression_values = 'scaled',
+ show_image = F,
+ feats = 'ERBB2')
13.6.4 Register post-Xenium H&E and IF image
For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics.
-HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
- imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
- micron = 0.2125)
-plot(HE_xen)
+HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
+ imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
+ micron = 0.2125)
+plot(HE_xen)
The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use.
-HE_xen <- HE_xen |> flip(direction = "vertical")
-gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
-plot(gimg_rast)
-#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
+HE_xen <- HE_xen |> flip(direction = "vertical")
+gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
+plot(gimg_rast)
+#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots
-gg <- ggplot2::ggplot()
-pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
-pl + ggplot2::geom_smooth() +
- ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+gg <- ggplot2::ggplot()
+pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
+pl + ggplot2::geom_smooth() +
+ ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.4.1 Add registered image information and compare RNA vs protein expression
With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF
-CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
-HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
-Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
- largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
+CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
+HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
+Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
+ largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
Get the cell polygons, as Xenium and IF are both subcellular resolution
-cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
-colnames(cellpoly_dt) <- c('poly_ID','x','y')
-cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
-Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
+cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
+colnames(cellpoly_dt) <- c('poly_ID','x','y')
+cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
+Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
Compute the gene expression matrix by overlay the cell polygons and giotto points.
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'cell')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "cell",
- feat_info = "rna",
- aggr_function = "sum")
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'rna',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'ERBB2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'cell')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "cell",
+ feat_info = "rna",
+ aggr_function = "sum")
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'rna',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'ERBB2',
+ point_size = 0.05)
Now we overlay the HER2 expression from the raster image with the cell polygons.
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'cell',
- image_names = c('HER2','CD20'))
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'HER2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'cell',
+ image_names = c('HER2','CD20'))
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'HER2',
+ point_size = 0.05)
We can also overlay the protein expression to Visium spots
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'visium',
- image_names = c('HER2','CD20'))
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "visium",
- feat_info = "protein",
- aggr_function = "sum")
-
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'visium',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'visium',
- feats = 'HER2',
- point_size = 2)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'visium',
+ image_names = c('HER2','CD20'))
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "visium",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'visium',
+ feats = 'HER2',
+ point_size = 2)
@@ -944,29 +944,29 @@ 13.6.4.1 Add registered image inf
13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation
Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially.
Here, we provide an example of two compressed images to show the automatic alignment pipeline.
-
+
-IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
-terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
+IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
+terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform.
-estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
- flip_horizontal = T,
- use_single_channel = T,
- single_channel_number = 3),
- y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
- invert = T),
- plot_match = T,
- max_ratio = 0.5,estimate_fun = 'Projective')
+estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
+ flip_horizontal = T,
+ use_single_channel = T,
+ single_channel_number = 3),
+ y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
+ invert = T),
+ plot_match = T,
+ max_ratio = 0.5,estimate_fun = 'Projective')
Use the estimation, we can quickly visualize the transformation
-mtx <- as.matrix(estimation$params)
-transformed <- affine(IF, mtx)
-To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
-plot(HE)
-plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
-SIFT_registration_overlay
+mtx <- as.matrix(estimation$params)
+transformed <- affine(IF, mtx)
+To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
+plot(HE)
+plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
+SIFT_registration_overlay
diff --git a/spatial-proteomics-multiplexed-immunofluorescence.html b/spatial-proteomics-multiplexed-immunofluorescence.html
index 684d4e8..f550962 100644
--- a/spatial-proteomics-multiplexed-immunofluorescence.html
+++ b/spatial-proteomics-multiplexed-immunofluorescence.html
@@ -518,33 +518,33 @@ 11 Spatial proteomics: Multiplexe
Junxiang Xu
August 6th 2024
Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so.
-#Install the Giotto Environment with Cellpose, note that we only need to do it once
-reticulate::conda_create(envname = "giotto_cellpose",
- python_version = 3.8)
-#.re.restartR()
-reticulate::use_condaenv('giotto_cellpose')
-reticulate::py_install(
- pip = TRUE,
- envname = 'giotto_cellpose',
- packages = c(
- "pandas",
- "networkx",
- "python-igraph",
- "leidenalg",
- "scikit-learn",
- "cellpose",
- "smfishhmrf",
- 'tifffile',
- 'scikit-image'
- )
-)
-#.rs.restartR()
+#Install the Giotto Environment with Cellpose, note that we only need to do it once
+reticulate::conda_create(envname = "giotto_cellpose",
+ python_version = 3.8)
+#.re.restartR()
+reticulate::use_condaenv('giotto_cellpose')
+reticulate::py_install(
+ pip = TRUE,
+ envname = 'giotto_cellpose',
+ packages = c(
+ "pandas",
+ "networkx",
+ "python-igraph",
+ "leidenalg",
+ "scikit-learn",
+ "cellpose",
+ "smfishhmrf",
+ 'tifffile',
+ 'scikit-image'
+ )
+)
+#.rs.restartR()
Now, activate the Giotto python environment.
-#.rs.restartR()
-# Activate the Giotto python environment of your choice
-GiottoClass::set_giotto_python_path('giotto_cellpose')
-# Check if cellpose was successfully installed
-GiottoUtils::package_check('cellpose',repository = 'pip')
+#.rs.restartR()
+# Activate the Giotto python environment of your choice
+GiottoClass::set_giotto_python_path('giotto_cellpose')
+# Check if cellpose was successfully installed
+GiottoUtils::package_check('cellpose',repository = 'pip')
11.1 Spatial Proteomics Technologies
This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix.
@@ -557,58 +557,58 @@
11.2 Raw data type coming out of
11.2.1 Use ome.tiff as an example output data to begin with
OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore.
-library(Giotto)
-instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
- save_plot = TRUE,
- show_plot = TRUE,
- python_path = 'giotto_cellpose')
-options(timeout = 9999999)
-download_dir =file.path(getwd(),'/data/02_session4/')
-destfile = file.path(download_dir,'Lunaphore.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
-list.files(paste0(download_dir,'/Lunaphore'))
+library(Giotto)
+instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ python_path = 'giotto_cellpose')
+options(timeout = 9999999)
+download_dir =file.path(getwd(),'/data/02_session4/')
+destfile = file.path(download_dir,'Lunaphore.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
+list.files(paste0(download_dir,'/Lunaphore'))
We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML.
-img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
-img_meta
+img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
+img_meta
However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it.
-##Get channel information
-Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
-node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
-channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
-channel_df
+##Get channel information
+Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
+node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
+channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
+channel_df
11.2.2 Use single channel images as an example output data to begin with
Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like.
-#Create multichannel raster and extract each single channels
-Luna_terra = terra::rast(Luna)
-names(Luna_terra) <- channel_df$Name
-gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
-plot(gimg_DAPI)
+#Create multichannel raster and extract each single channels
+Luna_terra = terra::rast(Luna)
+names(Luna_terra) <- channel_df$Name
+gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
+plot(gimg_DAPI)
Extract and save the raster image for future use.
-single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
-if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
-for (i in 1:nrow(channel_df)){
- single_channel <- terra::subset(Luna_terra, i)
- terra::writeRaster(single_channel,
- filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
- overwrite=T)
-}
+single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
+if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
+for (i in 1:nrow(channel_df)){
+ single_channel <- terra::subset(Luna_terra, i)
+ terra::writeRaster(single_channel,
+ filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
+ overwrite=T)
+}
Create a list of GiottoLargeImages using single channel rasters.
-file_names = list.files(single_channel_dir,full.names = TRUE)
-image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
-gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
- names = image_names,
- negative_y = F,
- flip_vertical = T)
-names(gimg_list) <- image_names
-plot(gimg_list[['Vimentin']])
+file_names = list.files(single_channel_dir,full.names = TRUE)
+image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
+gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
+ names = image_names,
+ negative_y = F,
+ flip_vertical = T)
+names(gimg_list) <- image_names
+plot(gimg_list[['Vimentin']])
@@ -618,34 +618,34 @@ 11.3 Cell Segmentation file to ge
11.3.1 Using segmentation output file from DeepCell(mesmer) as an example.
We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function.
-gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
-plot(gpoly_mesmer)
+gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
+plot(gpoly_mesmer)
We can also zoom in to check how does the segmentation look.
-zoom <- c(2000,2500,2000,2500)
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+zoom <- c(2000,2500,2000,2500)
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
11.3.2 Using Giotto wrapper of Cellpose to perform segmentation
-gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
-writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
-#Create a giotto image to evaluate segmentation
-gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
+gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
+writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
+#Create a giotto image to evaluate segmentation
+gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website
-doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
- mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- channel_1 = 0,
- channel_2 = 0,
- model_name = 'cyto3',
- batch_size=12)
-cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- shift_horizontal_step = F,
- shift_vertical_step = F,
- flip_vertical = T)
-plot(gimg_for_cellpose)
-plot(cpoly, add = T, border = 'red')
+doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
+ mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ channel_1 = 0,
+ channel_2 = 0,
+ model_name = 'cyto3',
+ batch_size=12)
+cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ shift_horizontal_step = F,
+ shift_vertical_step = F,
+ flip_vertical = T)
+plot(gimg_for_cellpose)
+plot(cpoly, add = T, border = 'red')
@@ -654,133 +654,133 @@ 11.4 Create a Giotto Object using
You will need to have
- list of giotto images
- giottoPolygon created from segmentation
-Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
- images = gimg_list,
- instructions = instrs)
-Lunaphore_giotto
+Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
+ images = gimg_list,
+ instructions = instrs)
+Lunaphore_giotto
11.4.1 Overlap to matrix
calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with
-Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
- spatial_info = 'cell',
- image_names = names(gimg_list))
-
-Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
- type ='intensity',
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-showGiottoExpression(Lunaphore_giotto)
+Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
+ spatial_info = 'cell',
+ image_names = names(gimg_list))
+
+Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
+ type ='intensity',
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+showGiottoExpression(Lunaphore_giotto)
11.4.2 Manipulate Expression information
For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile.
Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object
-expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
-filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
-Lunaphore_giotto <- setExpression(Lunaphore_giotto,
- feat_type = 'protein',
- x = createExprObj(filtered_expr_mtx),
- name = 'raw')
-showGiottoExpression(Lunaphore_giotto)
+expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
+filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
+Lunaphore_giotto <- setExpression(Lunaphore_giotto,
+ feat_type = 'protein',
+ x = createExprObj(filtered_expr_mtx),
+ name = 'raw')
+showGiottoExpression(Lunaphore_giotto)
11.4.3 Rescale polygons
rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis
-Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
- poly_info = 'cell',
- name = 'smallcell',
- fx = 0.7, fy = 0.7, calculate_centroids = T)
-smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
-plot(smallpoly,add = TRUE, border = "red",ext = zoom)
+Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
+ poly_info = 'cell',
+ name = 'smallcell',
+ fx = 0.7, fy = 0.7, calculate_centroids = T)
+smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+plot(smallpoly,add = TRUE, border = "red",ext = zoom)
11.4.4 Perform clustering and differential expression
The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering
-Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
-Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
-
-Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
- scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
- set_seed = TRUE)
-screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
+Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
+Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
+
+Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
+ scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
+ set_seed = TRUE)
+screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check.
-Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
- dimensions_to_use = 1:5,
- set_seed = TRUE)
-Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
-Lunaphore_giotto = doHclust(Lunaphore_giotto,
- k = 8,
- dim_reduction_to_use = 'cells',
- spat_unit = 'cell',
- feat_type = 'protein')
-
-spatInSituPlotPoints(gobject = Lunaphore_giotto,
- spat_unit = "cell",
- polygon_feat_type = "cell",
- show_polygon = T,
- feat_type = "protein",
- feats = NULL,
- polygon_fill = 'hclust',
- polygon_fill_as_factor = TRUE,
- polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
- polygon_color = 'black',
- background_color = "white")
+Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
+ dimensions_to_use = 1:5,
+ set_seed = TRUE)
+Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
+Lunaphore_giotto = doHclust(Lunaphore_giotto,
+ k = 8,
+ dim_reduction_to_use = 'cells',
+ spat_unit = 'cell',
+ feat_type = 'protein')
+
+spatInSituPlotPoints(gobject = Lunaphore_giotto,
+ spat_unit = "cell",
+ polygon_feat_type = "cell",
+ show_polygon = T,
+ feat_type = "protein",
+ feats = NULL,
+ polygon_fill = 'hclust',
+ polygon_fill_as_factor = TRUE,
+ polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
+ polygon_color = 'black',
+ background_color = "white")
Then we can check the heatmap of protein expression and determine the first round of cluster annotation
-cluster_column = 'hclust'
-plotMetaDataHeatmap(Lunaphore_giotto,
- spat_unit = 'cell',feat_type = 'protein',
- expression_values = "raw",
- metadata_cols = c(cluster_column),
- selected_feats = names(gimg_list),
- y_text_size = 8,
- show_values = 'zscores_rescaled')
+cluster_column = 'hclust'
+plotMetaDataHeatmap(Lunaphore_giotto,
+ spat_unit = 'cell',feat_type = 'protein',
+ expression_values = "raw",
+ metadata_cols = c(cluster_column),
+ selected_feats = names(gimg_list),
+ y_text_size = 8,
+ show_values = 'zscores_rescaled')
11.4.5 Give the cluster an annotation based on expression values
-
+
11.4.6 spatial network
This is to create a cellular neighborhood based on nearest neighbor of physical distance
-Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
-
-spatPlot2D(Lunaphore_giotto,
- show_network = T,
- network_color = 'blue',
- point_size = 1.5,
- cell_color = 'hclust')
+Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
+
+spatPlot2D(Lunaphore_giotto,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1.5,
+ cell_color = 'hclust')
11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions
This is using cellProximityEnrichment() to statistically identify cell type interactions
-cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
- cluster_column = 'cell_types',
- spatial_network_name = 'Delaunay_network',
- adjust_method = 'fdr',
- number_of_simulations = 2000)
-## barplot
-cellProximityBarplot(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- min_orig_ints = 5, min_sim_ints = 5)
+cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
+ cluster_column = 'cell_types',
+ spatial_network_name = 'Delaunay_network',
+ adjust_method = 'fdr',
+ number_of_simulations = 2000)
+## barplot
+cellProximityBarplot(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ min_orig_ints = 5, min_sim_ints = 5)
-## network
-cellProximityNetwork(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- remove_self_edges = T,
- only_show_enrichment_edges = F)
+## network
+cellProximityNetwork(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ remove_self_edges = T,
+ only_show_enrichment_edges = F)
diff --git a/visium-hd.html b/visium-hd.html
index 23aa210..f24cdd4 100644
--- a/visium-hd.html
+++ b/visium-hd.html
@@ -525,7 +525,7 @@ 9.1 Objective9.2 Background
9.2.1 Visium HD Technology
-
+
Figure 9.1: Overview of Visium HD. Source: 10X Genomics
@@ -536,7 +536,7 @@
9.2.1 Visium HD Technology
9.2.2 Colorectal Cancer Sample
-
+
Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics
@@ -550,7 +550,7 @@
9.2.2 Colorectal Cancer Sample9.3 Data Ingestion
9.3.1 Visium HD output data format
-
+
Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline.
@@ -560,19 +560,19 @@
9.3.1 Visium HD output data forma
9.4 Tiling and aggregation
The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells.
-
+
9.5 Scalability and projection functions
diff --git a/visium-part-i.html b/visium-part-i.html
index 5aa1658..c4746ee 100644
--- a/visium-part-i.html
+++ b/visium-part-i.html
@@ -520,7 +520,7 @@ 7 Visium Part I
7.1 The Visium technology
Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression.
-
+
Figure 7.1: Visum workflow. Source: 10X Genomics
@@ -536,73 +536,73 @@
7.2 Introduction to the spatial d
7.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/01_session5/")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
+dir.create("data/01_session5/")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”.
-
+
7.4 Create the Giotto object
createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/01_session5/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-## Provide the path to the visium folder
-data_path <- "data/01_session5/"
-
-## Create object directly from the visium folder
-visium_brain <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/01_session5/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+## Provide the path to the visium folder
+data_path <- "data/01_session5/"
+
+## Create object directly from the visium folder
+visium_brain <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
7.5 Subset on spots that were covered by tissue
Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area.
-spatPlot2D(
- gobject = visium_brain,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE)
-
+spatPlot2D(
+ gobject = visium_brain,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE)
+
Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area.
-
+
7.6 Quality control
@@ -610,43 +610,43 @@ 7.6 Quality controlvisium_brain_statistics <- addStatistics(gobject = visium_brain,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = visium_brain_statistics,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain_statistics <- addStatistics(gobject = visium_brain,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = visium_brain_statistics,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.3: Spatial distribution of features per spot.
filterDistributions() creates a histogram to show the distribution of features per spot across the sample.
-
-
+
+
Figure 7.4: Distribution of features per spot.
When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample.
-
-
+
+
Figure 7.5: Distribution of cells with different features per spot.
filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data:
-filterCombinations(gobject = visium_brain_statistics,
- expression_thresholds = c(1, 2, 3),
- feat_det_in_min_cells = c(50, 100, 200),
- min_det_feats_per_cell = c(500, 1000, 1500))
-
+filterCombinations(gobject = visium_brain_statistics,
+ expression_thresholds = c(1, 2, 3),
+ feat_det_in_min_cells = c(50, 100, 200),
+ min_det_feats_per_cell = c(500, 1000, 1500))
+
Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations.
@@ -656,34 +656,34 @@
7.6 Quality control
7.7 Filtering
Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample.
-visium_brain <- filterGiotto(
- gobject = visium_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE
-)
-Feature type: rna
-Number of cells removed: 4 out of 2702
-Number of feats removed: 7311 out of 22125
+
+
7.8 Normalization
Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one.
-visium_brain <- normalizeGiotto(
- gobject = visium_brain,
- scalefactor = 6000,
- verbose = TRUE
-)
+visium_brain <- normalizeGiotto(
+ gobject = visium_brain,
+ scalefactor = 6000,
+ verbose = TRUE
+)
Calculate the normalized number of features per spot and save the statistics in the metadata table.
-visium_brain <- addStatistics(gobject = visium_brain)
-
-## visualize
-spatPlot2D(gobject = visium_brain,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain <- addStatistics(gobject = visium_brain)
+
+## visualize
+spatPlot2D(gobject = visium_brain,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.7: Spatial distribution of the number of features per spot.
@@ -698,11 +698,11 @@
7.9.1 Highly Variable Features:
loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_loess",
- save_plot = TRUE,
- default_save_name = "HVFplot_loess")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_loess",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_loess")
+
Figure 7.8: Covariance of HVFs using the loess method.
@@ -711,11 +711,11 @@
7.9.1 Highly Variable Features:
pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")
+
Figure 7.9: Variance of HVFs using the pearson residuals method.
@@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics.
The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument.
-
+
14.2 Introduction to the spatial
Figure 14.2: CytAssist workflow. Source: 10X genomics. @@ -542,19 +542,19 @@
14.2 Introduction to the spatial
14.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/03_session3")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
+dir.create("data/03_session3")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”.
-
+
14.4 Create the Giotto object
@@ -564,124 +564,124 @@ 14.4 Create the Giotto objectx,y(,z) coordinates for cells or spots (or the path to)
createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/03_session3/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-# Provide the path to the data folder
-data_path <- "data/03_session3/"
-
-# Create object directly from the data folder
-visium_tonsil <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/03_session3/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+# Provide the path to the data folder
+data_path <- "data/03_session3/"
+
+# Create object directly from the data folder
+visium_tonsil <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
- Print the information of the object, note that both rna and protein are listed in the expression slot.
-
+
14.5 Subset on spots that were covered by tissue
-spatPlot2D(
- gobject = visium_tonsil,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE,
- image_name = "image"
-)
-
+spatPlot2D(
+ gobject = visium_tonsil,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE,
+ image_name = "image"
+)
+
Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object.
-
+
14.6 RNA processing
- Run the Filtering, normalization, and statistics steps using only the RNA feature.
-visium_tonsil <- filterGiotto(
- gobject = visium_tonsil,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil)
+visium_tonsil <- filterGiotto(
+ gobject = visium_tonsil,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1000,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil)
- Dimension reduction
Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs.
-visium_tonsil <- calculateHVF(gobject = visium_tonsil)
-
-visium_tonsil <- runPCA(gobject = visium_tonsil)
+visium_tonsil <- calculateHVF(gobject = visium_tonsil)
+
+visium_tonsil <- runPCA(gobject = visium_tonsil)
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the RNA-based Leiden clusters.
-
+
- Visualization
Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot.
-plotUMAP(gobject = visium_tonsil,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters.
Plot the spatial distribution of the RNA-based Leiden cluster per spot.
-spatPlot2D(gobject = visium_tonsil,
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.5: Spatial distribution of RNA-based Leiden clusters.
@@ -693,80 +693,80 @@
14.7 Protein processingvisium_tonsil <- filterGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein")
+visium_tonsil <- filterGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein")
- Dimension reduction
Calculate the principal components using all the proteins available in the dataset.
-
+
- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10,
- k = 30)
+visium_tonsil <- runUMAP(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10,
+ k = 30)
Calculate the Protein-based Leiden clusters.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- resolution = 1,
- n_iterations = 1000)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ resolution = 1,
+ n_iterations = 1000)
- Visualization
Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)
+
Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters.
Plot the spatial distribution of the Protein-based Leiden clusters.
-spatPlot2D(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
-
+spatPlot2D(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)
+
Figure 14.7: Spatial distribution of Protein-based Leiden clusters.
@@ -779,68 +779,68 @@
14.8 Multi-omics integrationCalculate kNN
Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type.
-## RNA modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
-
-## Protein modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
+## RNA modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
+
+## Protein modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
- Run WNN
Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object.
-visium_tonsil <- runWNN(visium_tonsil,
- spat_unit = "cell",
- modality_1 = "rna",
- modality_2 = "protein",
- pca_name_modality_1 = "pca",
- pca_name_modality_2 = "protein.pca",
- k = 20,
- integrated_feat_type = NULL,
- matrix_result_name = NULL,
- w_name_modality_1 = NULL,
- w_name_modality_2 = NULL,
- verbose = TRUE)
+visium_tonsil <- runWNN(visium_tonsil,
+ spat_unit = "cell",
+ modality_1 = "rna",
+ modality_2 = "protein",
+ pca_name_modality_1 = "pca",
+ pca_name_modality_2 = "protein.pca",
+ k = 20,
+ integrated_feat_type = NULL,
+ matrix_result_name = NULL,
+ w_name_modality_1 = NULL,
+ w_name_modality_2 = NULL,
+ verbose = TRUE)
- Run Integrated umap
Calculate the UMAP using the weights of each feature per spot.
-visium_tonsil <- runIntegratedUMAP(visium_tonsil,
- modality1 = "rna",
- modality2 = "protein",
- spread = 7,
- min_dist = 1,
- force = FALSE)
+visium_tonsil <- runIntegratedUMAP(visium_tonsil,
+ modality1 = "rna",
+ modality2 = "protein",
+ spread = 7,
+ min_dist = 1,
+ force = FALSE)
- Calculate integrated clusters
Calculate the multiomics-based Leiden clusters using the weights of each feature per spot.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- nn_network_to_use = "kNN",
- network_name = "integrated_kNN",
- name = "integrated_leiden_clus",
- resolution = 1)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ nn_network_to_use = "kNN",
+ network_name = "integrated_kNN",
+ name = "integrated_leiden_clus",
+ resolution = 1)
- Visualize the integrated UMAP
Plot the integrated UMAP and color the spots using the integrated Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- dim_reduction_name = "integrated.umap",
- point_size = 2,
- title = "Integrated UMAP using Integrated Leiden clusters")
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ dim_reduction_name = "integrated.umap",
+ point_size = 2,
+ title = "Integrated UMAP using Integrated Leiden clusters")
+
Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters.
@@ -850,14 +850,14 @@
14.8 Multi-omics integrationVisualize spatial plot with integrated clusters
Plot the spatial distribution of the integrated Leiden clusters.
-spatPlot2D(visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- point_size = 3,
- show_image = TRUE,
- title = "Integrated Leiden clustering")
-
+spatPlot2D(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ point_size = 3,
+ show_image = TRUE,
+ title = "Integrated Leiden clustering")
+
Figure 14.9: Spatial distribution of the integrated Leiden clusters.
@@ -866,85 +866,85 @@
14.8 Multi-omics integration
14.9 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
- [7] matrixStats_1.3.0 compiler_4.4.1
- [9] png_0.1-8 systemfonts_1.1.0
- [11] vctrs_0.6.5 reshape2_1.4.4
- [13] stringr_1.5.1 pkgconfig_2.0.3
- [15] SpatialExperiment_1.14.0 crayon_1.5.3
- [17] fastmap_1.2.0 backports_1.5.0
- [19] magick_2.8.4 XVector_0.44.0
- [21] labeling_0.4.3 utf8_1.2.4
- [23] rmarkdown_2.27 UCSC.utils_1.0.0
- [25] ragg_1.3.2 purrr_1.0.2
- [27] xfun_0.46 beachmat_2.20.0
- [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
- [31] jsonlite_1.8.8 DelayedArray_0.30.1
- [33] BiocParallel_1.38.0 terra_1.7-78
- [35] irlba_2.3.5.1 parallel_4.4.1
- [37] R6_2.5.1 stringi_1.8.4
- [39] RColorBrewer_1.1-3 reticulate_1.38.0
- [41] parallelly_1.37.1 GenomicRanges_1.56.1
- [43] scattermore_1.2 Rcpp_1.0.13
- [45] bookdown_0.40 SummarizedExperiment_1.34.0
- [47] knitr_1.48 future.apply_1.11.2
- [49] R.utils_2.12.3 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
+ [7] matrixStats_1.3.0 compiler_4.4.1
+ [9] png_0.1-8 systemfonts_1.1.0
+ [11] vctrs_0.6.5 reshape2_1.4.4
+ [13] stringr_1.5.1 pkgconfig_2.0.3
+ [15] SpatialExperiment_1.14.0 crayon_1.5.3
+ [17] fastmap_1.2.0 backports_1.5.0
+ [19] magick_2.8.4 XVector_0.44.0
+ [21] labeling_0.4.3 utf8_1.2.4
+ [23] rmarkdown_2.27 UCSC.utils_1.0.0
+ [25] ragg_1.3.2 purrr_1.0.2
+ [27] xfun_0.46 beachmat_2.20.0
+ [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
+ [31] jsonlite_1.8.8 DelayedArray_0.30.1
+ [33] BiocParallel_1.38.0 terra_1.7-78
+ [35] irlba_2.3.5.1 parallel_4.4.1
+ [37] R6_2.5.1 stringi_1.8.4
+ [39] RColorBrewer_1.1-3 reticulate_1.38.0
+ [41] parallelly_1.37.1 GenomicRanges_1.56.1
+ [43] scattermore_1.2 Rcpp_1.0.13
+ [45] bookdown_0.40 SummarizedExperiment_1.34.0
+ [47] knitr_1.48 future.apply_1.11.2
+ [49] R.utils_2.12.3 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/reference-keys.txt b/reference-keys.txt
index a17604d..555f956 100644
--- a/reference-keys.txt
+++ b/reference-keys.txt
@@ -4,115 +4,116 @@ fig:unnamed-chunk-14
fig:unnamed-chunk-15
fig:unnamed-chunk-16
fig:unnamed-chunk-18
-fig:unnamed-chunk-30
-fig:unnamed-chunk-35
-fig:unnamed-chunk-38
-fig:unnamed-chunk-40
+fig:unnamed-chunk-32
+fig:unnamed-chunk-37
fig:unnamed-chunk-42
-fig:unnamed-chunk-44
+fig:unnamed-chunk-45
+fig:unnamed-chunk-47
fig:unnamed-chunk-49
fig:unnamed-chunk-51
-fig:unnamed-chunk-53
-fig:unnamed-chunk-55
-fig:unnamed-chunk-59
-fig:unnamed-chunk-61
-fig:unnamed-chunk-63
+fig:unnamed-chunk-56
+fig:unnamed-chunk-58
+fig:unnamed-chunk-60
+fig:unnamed-chunk-62
fig:unnamed-chunk-66
-fig:unnamed-chunk-69
-fig:unnamed-chunk-74
+fig:unnamed-chunk-68
+fig:unnamed-chunk-70
+fig:unnamed-chunk-73
fig:unnamed-chunk-76
-fig:unnamed-chunk-78
-fig:unnamed-chunk-80
-fig:unnamed-chunk-82
-fig:unnamed-chunk-84
+fig:unnamed-chunk-81
+fig:unnamed-chunk-83
+fig:unnamed-chunk-85
fig:unnamed-chunk-87
+fig:unnamed-chunk-89
+fig:unnamed-chunk-91
fig:unnamed-chunk-94
-fig:unnamed-chunk-96
-fig:unnamed-chunk-98
fig:unnamed-chunk-101
fig:unnamed-chunk-103
fig:unnamed-chunk-105
+fig:unnamed-chunk-108
+fig:unnamed-chunk-110
fig:unnamed-chunk-112
-fig:unnamed-chunk-114
-fig:unnamed-chunk-118
-fig:unnamed-chunk-120
-fig:unnamed-chunk-123
-fig:unnamed-chunk-128
-fig:unnamed-chunk-131
-fig:unnamed-chunk-134
-fig:unnamed-chunk-137
+fig:unnamed-chunk-119
+fig:unnamed-chunk-121
+fig:unnamed-chunk-125
+fig:unnamed-chunk-127
+fig:unnamed-chunk-130
+fig:unnamed-chunk-135
+fig:unnamed-chunk-138
fig:unnamed-chunk-141
-fig:unnamed-chunk-143
-fig:unnamed-chunk-147
+fig:unnamed-chunk-144
+fig:unnamed-chunk-148
fig:unnamed-chunk-150
-fig:unnamed-chunk-152
+fig:unnamed-chunk-154
+fig:unnamed-chunk-157
fig:unnamed-chunk-159
-fig:unnamed-chunk-161
-fig:unnamed-chunk-163
-fig:unnamed-chunk-165
-fig:unnamed-chunk-167
-fig:unnamed-chunk-169
-fig:unnamed-chunk-171
+fig:unnamed-chunk-166
+fig:unnamed-chunk-168
+fig:unnamed-chunk-170
+fig:unnamed-chunk-172
fig:unnamed-chunk-174
-fig:unnamed-chunk-175
fig:unnamed-chunk-176
-fig:unnamed-chunk-185
-fig:unnamed-chunk-191
-fig:unnamed-chunk-194
+fig:unnamed-chunk-178
+fig:unnamed-chunk-181
+fig:unnamed-chunk-182
+fig:unnamed-chunk-183
+fig:unnamed-chunk-192
+fig:unnamed-chunk-198
fig:unnamed-chunk-201
-fig:unnamed-chunk-204
-fig:unnamed-chunk-206
-fig:unnamed-chunk-210
-fig:unnamed-chunk-212
-fig:unnamed-chunk-215
+fig:unnamed-chunk-208
+fig:unnamed-chunk-211
+fig:unnamed-chunk-213
fig:unnamed-chunk-217
fig:unnamed-chunk-219
-fig:unnamed-chunk-221
+fig:unnamed-chunk-222
fig:unnamed-chunk-224
-fig:unnamed-chunk-225
-fig:unnamed-chunk-227
-fig:unnamed-chunk-229
-fig:unnamed-chunk-230
+fig:unnamed-chunk-226
+fig:unnamed-chunk-228
+fig:unnamed-chunk-231
fig:unnamed-chunk-232
fig:unnamed-chunk-234
fig:unnamed-chunk-236
-fig:unnamed-chunk-275
-fig:unnamed-chunk-279
-fig:unnamed-chunk-281
-fig:unnamed-chunk-283
+fig:unnamed-chunk-237
+fig:unnamed-chunk-239
+fig:unnamed-chunk-241
+fig:unnamed-chunk-243
+fig:unnamed-chunk-282
fig:unnamed-chunk-286
fig:unnamed-chunk-288
fig:unnamed-chunk-290
-fig:unnamed-chunk-292
-fig:unnamed-chunk-294
+fig:unnamed-chunk-293
+fig:unnamed-chunk-295
fig:unnamed-chunk-297
fig:unnamed-chunk-299
fig:unnamed-chunk-301
-fig:unnamed-chunk-303
-fig:unnamed-chunk-305
-fig:unnamed-chunk-365
-fig:unnamed-chunk-366
+fig:unnamed-chunk-304
+fig:unnamed-chunk-306
+fig:unnamed-chunk-308
+fig:unnamed-chunk-310
+fig:unnamed-chunk-312
fig:unnamed-chunk-372
+fig:unnamed-chunk-373
fig:unnamed-chunk-379
-fig:unnamed-chunk-381
-fig:unnamed-chunk-387
-fig:unnamed-chunk-389
-fig:unnamed-chunk-395
-fig:unnamed-chunk-397
-fig:unnamed-chunk-462
-fig:unnamed-chunk-465
+fig:unnamed-chunk-386
+fig:unnamed-chunk-388
+fig:unnamed-chunk-394
+fig:unnamed-chunk-396
+fig:unnamed-chunk-402
+fig:unnamed-chunk-404
fig:unnamed-chunk-469
-fig:unnamed-chunk-470
fig:unnamed-chunk-472
-fig:unnamed-chunk-474
fig:unnamed-chunk-476
-fig:unnamed-chunk-478
+fig:unnamed-chunk-477
fig:unnamed-chunk-479
fig:unnamed-chunk-481
+fig:unnamed-chunk-483
fig:unnamed-chunk-485
+fig:unnamed-chunk-486
fig:unnamed-chunk-488
-fig:unnamed-chunk-490
fig:unnamed-chunk-492
+fig:unnamed-chunk-495
+fig:unnamed-chunk-497
+fig:unnamed-chunk-499
giotto-suite-workshop-2024
instructors
topics-and-schedule
@@ -153,6 +154,10 @@ presentation-1
ecosystem
installation-python-environment
giotto-installation-1
+system-prerequisites
+installation-of-released-version
+installation-of-dev-branch-giotto-packages
+common-install-issues
python-environment
default-installation
custom-installs
diff --git a/search_index.json b/search_index.json
index 2736898..72ab7c3 100644
--- a/search_index.json
+++ b/search_index.json
@@ -1 +1 @@
-[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: Ensure your system To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak. pak::pak("drieslab/Giotto") 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
+[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: 4.3.1.1 System prerequisites for windows, Rtools needs to be installed a major dependency terra needs GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux 4.3.1.2 Installation of released version To install the currently released version of Giotto in a single step: pak::pak("drieslab/Giotto") This should automatically install all the Giotto dependencies and other Giotto module packages. 4.3.1.3 Installation of dev branch Giotto packages You can install dev branch versions by using devtools::install_github() Core module dev branchs: “drieslab/Giotto@suite_dev” “drieslab/GiottoVisuals@dev” “drieslab/GiottoClass@dev” “drieslab/GiottoUtils@dev” devtools::install_github("drieslab/GiottoClass@dev") pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages. 4.3.1.4 Common install issues If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work. devtools::install_version("Matrix", version = "1.6-5") If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it. install.packages("irlba", type = "source") ## # Downloading packages ------------------------------------------------------- ## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s] ## Successfully downloaded 1 package in 1.2 seconds. ## ## The following package(s) will be installed: ## - irlba [2.3.5.1] ## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu". ## ## # Installing packages -------------------------------------------------------- ## - Installing irlba ... OK [built from source and cached in 5.7s] ## Successfully installed 1 package in 5.7 seconds. 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are: python_path - when set, will activate a python environment save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render. save_plot - whether to save plots to the save_dir return_plot - whether to return the plot objects. When FALSE, only NULL is returned show_plot - whether to show the plot in the viewer These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function. library(Giotto) save_dir <- "results/01_session2/" # this call will also intialize the python env instrs <- createGiottoInstructions( save_dir = save_dir, # working directory is the default show_plot = FALSE, save_plot = TRUE, return_plot = FALSE, python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default ) force(instrs) Giotto object creation functions all have an instructions param for passing in instructions objects. giotto objects will also respond to the instructions() generic. test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object # example plot g <- GiottoData::loadGiottoMini("visium") instructions(g) <- instrs instructions(g, "show_plot") # instructions say not to plot to viewer spatPlot2D(g, show_image = TRUE, image_name = "image") # instead it will directly write to the results folder As an example, you can also set individual instructions instructions(g, "show_plot") <- TRUE spatPlot2D(g, show_image = TRUE, image_name = "image") Figure 4.2: example image output "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
diff --git a/spatial-multi-modal-analysis.html b/spatial-multi-modal-analysis.html
index 5bdab74..c78f952 100644
--- a/spatial-multi-modal-analysis.html
+++ b/spatial-multi-modal-analysis.html
@@ -563,48 +563,48 @@ 13.2.3 Spatial classes:# load in data
-library(Giotto)
-
-g <- GiottoData::loadGiottoMini("vizgen")
-
-activeSpatUnit(g) <- "aggregate"
-
-gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE)
-
-gimg <- getGiottoImage(g)
+
13.3 Examples of the simple transforms with a giottoPolygon
-rain <- rainbow(nrow(gpoly))
-
-line_width <- 0.3
-
-# par to setup the grid plotting layout
-p <- par(no.readonly = TRUE)
-par(mfrow=c(3,3))
-gpoly |>
- plot(main = "no transform", col = rain, lwd = line_width)
-
-gpoly |> spatShift(dx = 1000) |>
- plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
-
-gpoly |> spin(45) |>
- plot(main = "spin(45)", col = rain, lwd = line_width)
-
-gpoly |> rescale(fx = 10, fy = 6) |>
- plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
-
-gpoly |> flip(direction = "vertical") |>
- plot(main = "flip()", col = rain, lwd = line_width)
-
-gpoly |> t() |>
- plot(main = "t()", col = rain, lwd = line_width)
-
-gpoly |> shear(fx = 0.5) |>
- plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
-par(p)
+rain <- rainbow(nrow(gpoly))
+
+line_width <- 0.3
+
+# par to setup the grid plotting layout
+p <- par(no.readonly = TRUE)
+par(mfrow=c(3,3))
+gpoly |>
+ plot(main = "no transform", col = rain, lwd = line_width)
+
+gpoly |> spatShift(dx = 1000) |>
+ plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
+
+gpoly |> spin(45) |>
+ plot(main = "spin(45)", col = rain, lwd = line_width)
+
+gpoly |> rescale(fx = 10, fy = 6) |>
+ plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
+
+gpoly |> flip(direction = "vertical") |>
+ plot(main = "flip()", col = rain, lwd = line_width)
+
+gpoly |> t() |>
+ plot(main = "t()", col = rain, lwd = line_width)
+
+gpoly |> shear(fx = 0.5) |>
+ plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
+par(p)
@@ -617,20 +617,20 @@ 13.4 Affine transforms# create affine2d
-aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
+# create affine2d
+aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set.
-
-# append several simple transforms
-aff <- aff |>
- spatShift(dx = 1000) |>
- spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- flip(direction = "vertical") |>
- t() |>
- shear(fx = 0.5)
-force(aff)
+
+# append several simple transforms
+aff <- aff |>
+ spatShift(dx = 1000) |>
+ spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ flip(direction = "vertical") |>
+ t() |>
+ shear(fx = 0.5)
+force(aff)
<affine2d>
anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax)
rotate : -0.785398163397448 (rad)
@@ -639,35 +639,35 @@ 13.4 Affine transformsplot(aff)
+
We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests.
-
+
13.5 Image transforms
Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting.
spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image.
-# affine transform of image info matches with polygon info
-gimg |> affine(aff) |> plot()
-gpoly |> affine(aff) |>
- plot(add = TRUE,
- border = "cyan",
- lwd = 0.3)
+# affine transform of image info matches with polygon info
+gimg |> affine(aff) |> plot()
+gpoly |> affine(aff) |>
+ plot(add = TRUE,
+ border = "cyan",
+ lwd = 0.3)
-# affine of the giotto object
-g |> affine(aff) |>
- spatInSituPlotPoints(
- show_image = TRUE,
- feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
- feats_color_code = rainbow(4),
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- point_size = 0.1,
- use_overlap = FALSE
- )
+# affine of the giotto object
+g |> affine(aff) |>
+ spatInSituPlotPoints(
+ show_image = TRUE,
+ feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
+ feats_color_code = rainbow(4),
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ point_size = 0.1,
+ use_overlap = FALSE
+ )
Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used.
@@ -687,256 +687,256 @@ 13.6.1 Example dataset: Xenium Br
– cooresponding centroid locations
The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance.
-library(Giotto)
-instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
- save_plot = TRUE,
- show_plot = TRUE)
-options(timeout = 999999)
-download_dir <-file.path(getwd(),'/data/03_session2/')
-destfile <- file.path(download_dir,'Multimodal_registration.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
-Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
-Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
+library(Giotto)
+instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
+ save_plot = TRUE,
+ show_plot = TRUE)
+options(timeout = 999999)
+download_dir <-file.path(getwd(),'/data/03_session2/')
+destfile <- file.path(download_dir,'Multimodal_registration.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
+Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
+Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')
13.6.2 Target Coordinate system
Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system.
-xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
-xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
-xen_cell_pl
+xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
+xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+xen_cell_pl
13.6.3 Visium to register
Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration
-G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
- gene_column_index = 2,
- png_name = 'tissue_hires_image.png',
- instructions = NULL)
-# In the meantime, calculate statistics for easier plot showing
-G_visium <- normalizeGiotto(G_visium)
-G_visium <- addStatistics(G_visium)
-V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
-V_origin
+G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
+ gene_column_index = 2,
+ png_name = 'tissue_hires_image.png',
+ instructions = NULL)
+# In the meantime, calculate statistics for easier plot showing
+G_visium <- normalizeGiotto(G_visium)
+G_visium <- addStatistics(G_visium)
+V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
+V_origin
The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform.
-# create affine2d
-aff <- affine(diag(c(1,1)))
-aff <- aff |>
- spin(90) |>
- flip(direction = "horizontal")
-force(aff)
-
-# Apply the transform
-V_tansformed <- affine(G_visium,aff)
-spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
-spatplot_to_register
+# create affine2d
+aff <- affine(diag(c(1,1)))
+aff <- aff |>
+ spin(90) |>
+ flip(direction = "horizontal")
+force(aff)
+
+# Apply the transform
+V_tansformed <- affine(G_visium,aff)
+spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
+spatplot_to_register
Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images.
Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app.
-
+
Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result.
-landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
-affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
-
-V_final <- affine(G_visium,affine_mtx %*% aff@affine)
-
-spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
-spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
+affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
+
+V_final <- affine(G_visium,affine_mtx %*% aff@affine)
+
+spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
+spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.3.1 Create Pseudo Visium dataset for comparison
Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance.
-V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
-spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
-spatPlot2D(V_final,
- show_network = T,
- network_color = 'blue',
- point_size = 1)
-center_to_center <- min(spat_network$distance)
-radius <- center_to_center*55/200
+V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
+spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
+spatPlot2D(V_final,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1)
+center_to_center <- min(spat_network$distance)
+radius <- center_to_center*55/200
Now we get the Pseudo Visium polygons
-Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
-stamp_dt <- circleVertices(radius = radius, npoints = 100)
-pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
-pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
-plot(pseudo_visium_poly)
+Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
+stamp_dt <- circleVertices(radius = radius, npoints = 100)
+pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
+pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
+plot(pseudo_visium_poly)
Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points
-xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
-gpoints <- createGiottoPoints(xen_transcripts)
-Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
- gpolygons = list('visium' = pseudo_visium_poly))
+xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
+gpoints <- createGiottoPoints(xen_transcripts)
+Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
+ gpolygons = list('visium' = pseudo_visium_poly))
Get gene expression information by overlapping polygon to points
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'visium')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "visium",
- feat_info = "rna",
- aggr_function = "sum")
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'visium')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "visium",
+ feat_info = "rna",
+ aggr_function = "sum")
Manipulate the expression for plotting
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'visium',
- name = 'plot')
-
-spatFeatPlot2D(Xen_obj,
- point_size = 3.5,
- expression_values = 'plot',
- show_image = F,
- feats = 'ERBB2')
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ name = 'plot')
+
+spatFeatPlot2D(Xen_obj,
+ point_size = 3.5,
+ expression_values = 'plot',
+ show_image = F,
+ feats = 'ERBB2')
Subset the registered Visium and plot same gene
-#get the extent of giotto points, xmin, xmax, ymin, ymax
-subset_extent <- ext(gpoints@spatVector)
-sub_visium <- subsetGiottoLocs(V_final,
- x_min = subset_extent[1],
- x_max = subset_extent[2],
- y_min = subset_extent[3],
- y_max = subset_extent[4])
-spatFeatPlot2D(sub_visium,
- point_size = 2,
- expression_values = 'scaled',
- show_image = F,
- feats = 'ERBB2')
+#get the extent of giotto points, xmin, xmax, ymin, ymax
+subset_extent <- ext(gpoints@spatVector)
+sub_visium <- subsetGiottoLocs(V_final,
+ x_min = subset_extent[1],
+ x_max = subset_extent[2],
+ y_min = subset_extent[3],
+ y_max = subset_extent[4])
+spatFeatPlot2D(sub_visium,
+ point_size = 2,
+ expression_values = 'scaled',
+ show_image = F,
+ feats = 'ERBB2')
13.6.4 Register post-Xenium H&E and IF image
For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics.
-HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
- imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
- micron = 0.2125)
-plot(HE_xen)
+HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
+ imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
+ micron = 0.2125)
+plot(HE_xen)
The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use.
-HE_xen <- HE_xen |> flip(direction = "vertical")
-gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
-plot(gimg_rast)
-#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
+HE_xen <- HE_xen |> flip(direction = "vertical")
+gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
+plot(gimg_rast)
+#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots
-gg <- ggplot2::ggplot()
-pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
-pl + ggplot2::geom_smooth() +
- ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+gg <- ggplot2::ggplot()
+pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
+pl + ggplot2::geom_smooth() +
+ ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.4.1 Add registered image information and compare RNA vs protein expression
With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF
-CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
-HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
-Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
- largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
+CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
+HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
+Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
+ largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))
Get the cell polygons, as Xenium and IF are both subcellular resolution
-cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
-colnames(cellpoly_dt) <- c('poly_ID','x','y')
-cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
-Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
+cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
+colnames(cellpoly_dt) <- c('poly_ID','x','y')
+cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
+Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))
Compute the gene expression matrix by overlay the cell polygons and giotto points.
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'cell')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "cell",
- feat_info = "rna",
- aggr_function = "sum")
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'rna',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'ERBB2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'cell')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "cell",
+ feat_info = "rna",
+ aggr_function = "sum")
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'rna',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'ERBB2',
+ point_size = 0.05)
Now we overlay the HER2 expression from the raster image with the cell polygons.
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'cell',
- image_names = c('HER2','CD20'))
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'HER2',
- point_size = 0.05)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'cell',
+ image_names = c('HER2','CD20'))
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'HER2',
+ point_size = 0.05)
We can also overlay the protein expression to Visium spots
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'visium',
- image_names = c('HER2','CD20'))
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "visium",
- feat_info = "protein",
- aggr_function = "sum")
-
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'visium',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'visium',
- feats = 'HER2',
- point_size = 2)
+Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'visium',
+ image_names = c('HER2','CD20'))
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "visium",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'visium',
+ feats = 'HER2',
+ point_size = 2)
@@ -944,29 +944,29 @@ 13.6.4.1 Add registered image inf
13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation
Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially.
Here, we provide an example of two compressed images to show the automatic alignment pipeline.
-
+
-IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
-terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
+IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
+terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform.
-estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
- flip_horizontal = T,
- use_single_channel = T,
- single_channel_number = 3),
- y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
- invert = T),
- plot_match = T,
- max_ratio = 0.5,estimate_fun = 'Projective')
+estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
+ flip_horizontal = T,
+ use_single_channel = T,
+ single_channel_number = 3),
+ y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
+ invert = T),
+ plot_match = T,
+ max_ratio = 0.5,estimate_fun = 'Projective')
Use the estimation, we can quickly visualize the transformation
-mtx <- as.matrix(estimation$params)
-transformed <- affine(IF, mtx)
-To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
-plot(HE)
-plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
-SIFT_registration_overlay
+mtx <- as.matrix(estimation$params)
+transformed <- affine(IF, mtx)
+To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
+plot(HE)
+plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
+SIFT_registration_overlay
diff --git a/spatial-proteomics-multiplexed-immunofluorescence.html b/spatial-proteomics-multiplexed-immunofluorescence.html
index 684d4e8..f550962 100644
--- a/spatial-proteomics-multiplexed-immunofluorescence.html
+++ b/spatial-proteomics-multiplexed-immunofluorescence.html
@@ -518,33 +518,33 @@ 11 Spatial proteomics: Multiplexe
Junxiang Xu
August 6th 2024
Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so.
-#Install the Giotto Environment with Cellpose, note that we only need to do it once
-reticulate::conda_create(envname = "giotto_cellpose",
- python_version = 3.8)
-#.re.restartR()
-reticulate::use_condaenv('giotto_cellpose')
-reticulate::py_install(
- pip = TRUE,
- envname = 'giotto_cellpose',
- packages = c(
- "pandas",
- "networkx",
- "python-igraph",
- "leidenalg",
- "scikit-learn",
- "cellpose",
- "smfishhmrf",
- 'tifffile',
- 'scikit-image'
- )
-)
-#.rs.restartR()
+#Install the Giotto Environment with Cellpose, note that we only need to do it once
+reticulate::conda_create(envname = "giotto_cellpose",
+ python_version = 3.8)
+#.re.restartR()
+reticulate::use_condaenv('giotto_cellpose')
+reticulate::py_install(
+ pip = TRUE,
+ envname = 'giotto_cellpose',
+ packages = c(
+ "pandas",
+ "networkx",
+ "python-igraph",
+ "leidenalg",
+ "scikit-learn",
+ "cellpose",
+ "smfishhmrf",
+ 'tifffile',
+ 'scikit-image'
+ )
+)
+#.rs.restartR()
Now, activate the Giotto python environment.
-#.rs.restartR()
-# Activate the Giotto python environment of your choice
-GiottoClass::set_giotto_python_path('giotto_cellpose')
-# Check if cellpose was successfully installed
-GiottoUtils::package_check('cellpose',repository = 'pip')
+#.rs.restartR()
+# Activate the Giotto python environment of your choice
+GiottoClass::set_giotto_python_path('giotto_cellpose')
+# Check if cellpose was successfully installed
+GiottoUtils::package_check('cellpose',repository = 'pip')
11.1 Spatial Proteomics Technologies
This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix.
@@ -557,58 +557,58 @@
11.2 Raw data type coming out of
11.2.1 Use ome.tiff as an example output data to begin with
OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore.
-library(Giotto)
-instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
- save_plot = TRUE,
- show_plot = TRUE,
- python_path = 'giotto_cellpose')
-options(timeout = 9999999)
-download_dir =file.path(getwd(),'/data/02_session4/')
-destfile = file.path(download_dir,'Lunaphore.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
-list.files(paste0(download_dir,'/Lunaphore'))
+library(Giotto)
+instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ python_path = 'giotto_cellpose')
+options(timeout = 9999999)
+download_dir =file.path(getwd(),'/data/02_session4/')
+destfile = file.path(download_dir,'Lunaphore.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
+list.files(paste0(download_dir,'/Lunaphore'))
We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML.
-img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
-img_meta
+img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
+img_meta
However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it.
-##Get channel information
-Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
-node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
-channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
-channel_df
+##Get channel information
+Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
+node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
+channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
+channel_df
11.2.2 Use single channel images as an example output data to begin with
Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like.
-#Create multichannel raster and extract each single channels
-Luna_terra = terra::rast(Luna)
-names(Luna_terra) <- channel_df$Name
-gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
-plot(gimg_DAPI)
+#Create multichannel raster and extract each single channels
+Luna_terra = terra::rast(Luna)
+names(Luna_terra) <- channel_df$Name
+gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
+plot(gimg_DAPI)
Extract and save the raster image for future use.
-single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
-if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
-for (i in 1:nrow(channel_df)){
- single_channel <- terra::subset(Luna_terra, i)
- terra::writeRaster(single_channel,
- filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
- overwrite=T)
-}
+single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
+if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
+for (i in 1:nrow(channel_df)){
+ single_channel <- terra::subset(Luna_terra, i)
+ terra::writeRaster(single_channel,
+ filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
+ overwrite=T)
+}
Create a list of GiottoLargeImages using single channel rasters.
-file_names = list.files(single_channel_dir,full.names = TRUE)
-image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
-gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
- names = image_names,
- negative_y = F,
- flip_vertical = T)
-names(gimg_list) <- image_names
-plot(gimg_list[['Vimentin']])
+file_names = list.files(single_channel_dir,full.names = TRUE)
+image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
+gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
+ names = image_names,
+ negative_y = F,
+ flip_vertical = T)
+names(gimg_list) <- image_names
+plot(gimg_list[['Vimentin']])
@@ -618,34 +618,34 @@ 11.3 Cell Segmentation file to ge
11.3.1 Using segmentation output file from DeepCell(mesmer) as an example.
We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function.
-gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
-plot(gpoly_mesmer)
+gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
+plot(gpoly_mesmer)
We can also zoom in to check how does the segmentation look.
-zoom <- c(2000,2500,2000,2500)
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+zoom <- c(2000,2500,2000,2500)
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
11.3.2 Using Giotto wrapper of Cellpose to perform segmentation
-gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
-writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
-#Create a giotto image to evaluate segmentation
-gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
+gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
+writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
+#Create a giotto image to evaluate segmentation
+gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website
-doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
- mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- channel_1 = 0,
- channel_2 = 0,
- model_name = 'cyto3',
- batch_size=12)
-cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- shift_horizontal_step = F,
- shift_vertical_step = F,
- flip_vertical = T)
-plot(gimg_for_cellpose)
-plot(cpoly, add = T, border = 'red')
+doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
+ mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ channel_1 = 0,
+ channel_2 = 0,
+ model_name = 'cyto3',
+ batch_size=12)
+cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ shift_horizontal_step = F,
+ shift_vertical_step = F,
+ flip_vertical = T)
+plot(gimg_for_cellpose)
+plot(cpoly, add = T, border = 'red')
@@ -654,133 +654,133 @@ 11.4 Create a Giotto Object using
You will need to have
- list of giotto images
- giottoPolygon created from segmentation
-Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
- images = gimg_list,
- instructions = instrs)
-Lunaphore_giotto
+Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
+ images = gimg_list,
+ instructions = instrs)
+Lunaphore_giotto
11.4.1 Overlap to matrix
calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with
-Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
- spatial_info = 'cell',
- image_names = names(gimg_list))
-
-Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
- type ='intensity',
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-showGiottoExpression(Lunaphore_giotto)
+Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
+ spatial_info = 'cell',
+ image_names = names(gimg_list))
+
+Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
+ type ='intensity',
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+showGiottoExpression(Lunaphore_giotto)
11.4.2 Manipulate Expression information
For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile.
Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object
-expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
-filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
-Lunaphore_giotto <- setExpression(Lunaphore_giotto,
- feat_type = 'protein',
- x = createExprObj(filtered_expr_mtx),
- name = 'raw')
-showGiottoExpression(Lunaphore_giotto)
+expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
+filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
+Lunaphore_giotto <- setExpression(Lunaphore_giotto,
+ feat_type = 'protein',
+ x = createExprObj(filtered_expr_mtx),
+ name = 'raw')
+showGiottoExpression(Lunaphore_giotto)
11.4.3 Rescale polygons
rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis
-Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
- poly_info = 'cell',
- name = 'smallcell',
- fx = 0.7, fy = 0.7, calculate_centroids = T)
-smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
-plot(smallpoly,add = TRUE, border = "red",ext = zoom)
+Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
+ poly_info = 'cell',
+ name = 'smallcell',
+ fx = 0.7, fy = 0.7, calculate_centroids = T)
+smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+plot(smallpoly,add = TRUE, border = "red",ext = zoom)
11.4.4 Perform clustering and differential expression
The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering
-Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
-Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
-
-Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
- scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
- set_seed = TRUE)
-screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
+Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
+Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
+
+Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
+ scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
+ set_seed = TRUE)
+screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check.
-Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
- dimensions_to_use = 1:5,
- set_seed = TRUE)
-Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
-Lunaphore_giotto = doHclust(Lunaphore_giotto,
- k = 8,
- dim_reduction_to_use = 'cells',
- spat_unit = 'cell',
- feat_type = 'protein')
-
-spatInSituPlotPoints(gobject = Lunaphore_giotto,
- spat_unit = "cell",
- polygon_feat_type = "cell",
- show_polygon = T,
- feat_type = "protein",
- feats = NULL,
- polygon_fill = 'hclust',
- polygon_fill_as_factor = TRUE,
- polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
- polygon_color = 'black',
- background_color = "white")
+Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
+ dimensions_to_use = 1:5,
+ set_seed = TRUE)
+Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
+Lunaphore_giotto = doHclust(Lunaphore_giotto,
+ k = 8,
+ dim_reduction_to_use = 'cells',
+ spat_unit = 'cell',
+ feat_type = 'protein')
+
+spatInSituPlotPoints(gobject = Lunaphore_giotto,
+ spat_unit = "cell",
+ polygon_feat_type = "cell",
+ show_polygon = T,
+ feat_type = "protein",
+ feats = NULL,
+ polygon_fill = 'hclust',
+ polygon_fill_as_factor = TRUE,
+ polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
+ polygon_color = 'black',
+ background_color = "white")
Then we can check the heatmap of protein expression and determine the first round of cluster annotation
-cluster_column = 'hclust'
-plotMetaDataHeatmap(Lunaphore_giotto,
- spat_unit = 'cell',feat_type = 'protein',
- expression_values = "raw",
- metadata_cols = c(cluster_column),
- selected_feats = names(gimg_list),
- y_text_size = 8,
- show_values = 'zscores_rescaled')
+cluster_column = 'hclust'
+plotMetaDataHeatmap(Lunaphore_giotto,
+ spat_unit = 'cell',feat_type = 'protein',
+ expression_values = "raw",
+ metadata_cols = c(cluster_column),
+ selected_feats = names(gimg_list),
+ y_text_size = 8,
+ show_values = 'zscores_rescaled')
11.4.5 Give the cluster an annotation based on expression values
-
+
11.4.6 spatial network
This is to create a cellular neighborhood based on nearest neighbor of physical distance
-Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
-
-spatPlot2D(Lunaphore_giotto,
- show_network = T,
- network_color = 'blue',
- point_size = 1.5,
- cell_color = 'hclust')
+Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
+
+spatPlot2D(Lunaphore_giotto,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1.5,
+ cell_color = 'hclust')
11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions
This is using cellProximityEnrichment() to statistically identify cell type interactions
-cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
- cluster_column = 'cell_types',
- spatial_network_name = 'Delaunay_network',
- adjust_method = 'fdr',
- number_of_simulations = 2000)
-## barplot
-cellProximityBarplot(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- min_orig_ints = 5, min_sim_ints = 5)
+cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
+ cluster_column = 'cell_types',
+ spatial_network_name = 'Delaunay_network',
+ adjust_method = 'fdr',
+ number_of_simulations = 2000)
+## barplot
+cellProximityBarplot(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ min_orig_ints = 5, min_sim_ints = 5)
-## network
-cellProximityNetwork(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- remove_self_edges = T,
- only_show_enrichment_edges = F)
+## network
+cellProximityNetwork(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ remove_self_edges = T,
+ only_show_enrichment_edges = F)
diff --git a/visium-hd.html b/visium-hd.html
index 23aa210..f24cdd4 100644
--- a/visium-hd.html
+++ b/visium-hd.html
@@ -525,7 +525,7 @@ 9.1 Objective9.2 Background
9.2.1 Visium HD Technology
-
+
Figure 9.1: Overview of Visium HD. Source: 10X Genomics
@@ -536,7 +536,7 @@
9.2.1 Visium HD Technology
9.2.2 Colorectal Cancer Sample
-
+
Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics
@@ -550,7 +550,7 @@
9.2.2 Colorectal Cancer Sample9.3 Data Ingestion
9.3.1 Visium HD output data format
-
+
Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline.
@@ -560,19 +560,19 @@
9.3.1 Visium HD output data forma
9.4 Tiling and aggregation
The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells.
-
+
9.5 Scalability and projection functions
diff --git a/visium-part-i.html b/visium-part-i.html
index 5aa1658..c4746ee 100644
--- a/visium-part-i.html
+++ b/visium-part-i.html
@@ -520,7 +520,7 @@ 7 Visium Part I
7.1 The Visium technology
Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression.
-
+
Figure 7.1: Visum workflow. Source: 10X Genomics
@@ -536,73 +536,73 @@
7.2 Introduction to the spatial d
7.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/01_session5/")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
+dir.create("data/01_session5/")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”.
-
+
7.4 Create the Giotto object
createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/01_session5/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-## Provide the path to the visium folder
-data_path <- "data/01_session5/"
-
-## Create object directly from the visium folder
-visium_brain <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/01_session5/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+## Provide the path to the visium folder
+data_path <- "data/01_session5/"
+
+## Create object directly from the visium folder
+visium_brain <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
7.5 Subset on spots that were covered by tissue
Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area.
-spatPlot2D(
- gobject = visium_brain,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE)
-
+spatPlot2D(
+ gobject = visium_brain,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE)
+
Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area.
-
+
7.6 Quality control
@@ -610,43 +610,43 @@ 7.6 Quality controlvisium_brain_statistics <- addStatistics(gobject = visium_brain,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = visium_brain_statistics,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain_statistics <- addStatistics(gobject = visium_brain,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = visium_brain_statistics,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.3: Spatial distribution of features per spot.
filterDistributions() creates a histogram to show the distribution of features per spot across the sample.
-
-
+
+
Figure 7.4: Distribution of features per spot.
When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample.
-
-
+
+
Figure 7.5: Distribution of cells with different features per spot.
filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data:
-filterCombinations(gobject = visium_brain_statistics,
- expression_thresholds = c(1, 2, 3),
- feat_det_in_min_cells = c(50, 100, 200),
- min_det_feats_per_cell = c(500, 1000, 1500))
-
+filterCombinations(gobject = visium_brain_statistics,
+ expression_thresholds = c(1, 2, 3),
+ feat_det_in_min_cells = c(50, 100, 200),
+ min_det_feats_per_cell = c(500, 1000, 1500))
+
Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations.
@@ -656,34 +656,34 @@
7.6 Quality control
7.7 Filtering
Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample.
-visium_brain <- filterGiotto(
- gobject = visium_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE
-)
-Feature type: rna
-Number of cells removed: 4 out of 2702
-Number of feats removed: 7311 out of 22125
+
+
7.8 Normalization
Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one.
-visium_brain <- normalizeGiotto(
- gobject = visium_brain,
- scalefactor = 6000,
- verbose = TRUE
-)
+visium_brain <- normalizeGiotto(
+ gobject = visium_brain,
+ scalefactor = 6000,
+ verbose = TRUE
+)
Calculate the normalized number of features per spot and save the statistics in the metadata table.
-visium_brain <- addStatistics(gobject = visium_brain)
-
-## visualize
-spatPlot2D(gobject = visium_brain,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain <- addStatistics(gobject = visium_brain)
+
+## visualize
+spatPlot2D(gobject = visium_brain,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.7: Spatial distribution of the number of features per spot.
@@ -698,11 +698,11 @@
7.9.1 Highly Variable Features:
loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_loess",
- save_plot = TRUE,
- default_save_name = "HVFplot_loess")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_loess",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_loess")
+
Figure 7.8: Covariance of HVFs using the loess method.
@@ -711,11 +711,11 @@
7.9.1 Highly Variable Features:
pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")
+
Figure 7.9: Variance of HVFs using the pearson residuals method.
@@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
14.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/03_session3")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
- destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")dir.create("data/03_session3")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz",
+ destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz")After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”.
- +14.4 Create the Giotto object
@@ -564,124 +564,124 @@14.4 Create the Giotto objectx,y(,z) coordinates for cells or spots (or the path to)
createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/03_session3/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-# Provide the path to the data folder
-data_path <- "data/03_session3/"
-
-# Create object directly from the data folder
-visium_tonsil <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/03_session3/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+# Provide the path to the data folder
+data_path <- "data/03_session3/"
+
+# Create object directly from the data folder
+visium_tonsil <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
- Print the information of the object, note that both rna and protein are listed in the expression slot.
-
+
library(Giotto)
-
-## Set instructions
-results_folder <- "results/03_session3/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-# Provide the path to the data folder
-data_path <- "data/03_session3/"
-
-# Create object directly from the data folder
-visium_tonsil <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)library(Giotto)
+
+## Set instructions
+results_folder <- "results/03_session3/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+# Provide the path to the data folder
+data_path <- "data/03_session3/"
+
+# Create object directly from the data folder
+visium_tonsil <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)14.5 Subset on spots that were covered by tissue
-spatPlot2D(
- gobject = visium_tonsil,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE,
- image_name = "image"
-)
+
+
spatPlot2D(
+ gobject = visium_tonsil,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE,
+ image_name = "image"
+)Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object.
- +14.6 RNA processing
- Run the Filtering, normalization, and statistics steps using only the RNA feature.
visium_tonsil <- filterGiotto(
- gobject = visium_tonsil,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil)visium_tonsil <- filterGiotto(
+ gobject = visium_tonsil,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1000,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil)- Dimension reduction
Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs.
-visium_tonsil <- calculateHVF(gobject = visium_tonsil)
-
-visium_tonsil <- runPCA(gobject = visium_tonsil)visium_tonsil <- calculateHVF(gobject = visium_tonsil)
+
+visium_tonsil <- runPCA(gobject = visium_tonsil)- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- dimensions_to_use = 1:10,
- k = 30)visium_tonsil <- runUMAP(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ dimensions_to_use = 1:10,
+ k = 30)Calculate the RNA-based Leiden clusters.
- +- Visualization
Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot.
-plotUMAP(gobject = visium_tonsil,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
+
+
-
+
+
+
+
+
+
+
+
+
+
diff --git a/visium-hd.html b/visium-hd.html
index 23aa210..f24cdd4 100644
--- a/visium-hd.html
+++ b/visium-hd.html
@@ -525,7 +525,7 @@
plotUMAP(gobject = visium_tonsil,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters.
Plot the spatial distribution of the RNA-based Leiden cluster per spot.
-spatPlot2D(gobject = visium_tonsil,
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
+
+
14.7 Protein processing
+
+
+
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
@@ -944,29 +944,29 @@ 13.6.4.1 Add registered image inf
+
+
+
spatPlot2D(gobject = visium_tonsil,
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)Figure 14.5: Spatial distribution of RNA-based Leiden clusters. @@ -693,80 +693,80 @@
14.7 Protein processingvisium_tonsil <- filterGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein")
visium_tonsil <- filterGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1,
- expression_values = "raw",
- verbose = TRUE)
-
-visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- scalefactor = 6000,
- verbose = TRUE)
-
-visium_tonsil <- addStatistics(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein")visium_tonsil <- filterGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 1,
+ expression_values = "raw",
+ verbose = TRUE)
+
+visium_tonsil <- normalizeGiotto(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ scalefactor = 6000,
+ verbose = TRUE)
+
+visium_tonsil <- addStatistics(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein")- Dimension reduction
Calculate the principal components using all the proteins available in the dataset.
- +- Clustering
Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality.
-visium_tonsil <- runUMAP(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- runtSNE(visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10)
-
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- dimensions_to_use = 1:10,
- k = 30)visium_tonsil <- runUMAP(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- runtSNE(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10)
+
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ dimensions_to_use = 1:10,
+ k = 30)Calculate the Protein-based Leiden clusters.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- resolution = 1,
- n_iterations = 1000)visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ resolution = 1,
+ n_iterations = 1000)- Visualization
Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2)
+
+
-
+
+
@@ -687,256 +687,256 @@
+
plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2)Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters.
Plot the spatial distribution of the Protein-based Leiden clusters.
-spatPlot2D(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- show_image = TRUE,
- cell_color = "leiden_clus",
- point_size = 3)
+
+
14.8 Multi-omics integrationCalculate kNN
+
spatPlot2D(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ show_image = TRUE,
+ cell_color = "leiden_clus",
+ point_size = 3)Figure 14.7: Spatial distribution of Protein-based Leiden clusters. @@ -779,68 +779,68 @@
14.8 Multi-omics integrationCalculate kNN
Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type.
-## RNA modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
-
-## Protein modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
+## RNA modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
+
+## Protein modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
- Run WNN
Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object.
-visium_tonsil <- runWNN(visium_tonsil,
- spat_unit = "cell",
- modality_1 = "rna",
- modality_2 = "protein",
- pca_name_modality_1 = "pca",
- pca_name_modality_2 = "protein.pca",
- k = 20,
- integrated_feat_type = NULL,
- matrix_result_name = NULL,
- w_name_modality_1 = NULL,
- w_name_modality_2 = NULL,
- verbose = TRUE)
+visium_tonsil <- runWNN(visium_tonsil,
+ spat_unit = "cell",
+ modality_1 = "rna",
+ modality_2 = "protein",
+ pca_name_modality_1 = "pca",
+ pca_name_modality_2 = "protein.pca",
+ k = 20,
+ integrated_feat_type = NULL,
+ matrix_result_name = NULL,
+ w_name_modality_1 = NULL,
+ w_name_modality_2 = NULL,
+ verbose = TRUE)
- Run Integrated umap
Calculate the UMAP using the weights of each feature per spot.
-visium_tonsil <- runIntegratedUMAP(visium_tonsil,
- modality1 = "rna",
- modality2 = "protein",
- spread = 7,
- min_dist = 1,
- force = FALSE)
+visium_tonsil <- runIntegratedUMAP(visium_tonsil,
+ modality1 = "rna",
+ modality2 = "protein",
+ spread = 7,
+ min_dist = 1,
+ force = FALSE)
- Calculate integrated clusters
Calculate the multiomics-based Leiden clusters using the weights of each feature per spot.
-visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- nn_network_to_use = "kNN",
- network_name = "integrated_kNN",
- name = "integrated_leiden_clus",
- resolution = 1)
+visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ nn_network_to_use = "kNN",
+ network_name = "integrated_kNN",
+ name = "integrated_leiden_clus",
+ resolution = 1)
- Visualize the integrated UMAP
Plot the integrated UMAP and color the spots using the integrated Leiden clusters.
-plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- dim_reduction_name = "integrated.umap",
- point_size = 2,
- title = "Integrated UMAP using Integrated Leiden clusters")
-
+plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ dim_reduction_name = "integrated.umap",
+ point_size = 2,
+ title = "Integrated UMAP using Integrated Leiden clusters")
+
Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters.
@@ -850,14 +850,14 @@
14.8 Multi-omics integrationVisualize spatial plot with integrated clusters
Plot the spatial distribution of the integrated Leiden clusters.
-spatPlot2D(visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- point_size = 3,
- show_image = TRUE,
- title = "Integrated Leiden clustering")
-
+spatPlot2D(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ point_size = 3,
+ show_image = TRUE,
+ title = "Integrated Leiden clustering")
+
Figure 14.9: Spatial distribution of the integrated Leiden clusters.
@@ -866,85 +866,85 @@
14.8 Multi-omics integration
14.9 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
- [7] matrixStats_1.3.0 compiler_4.4.1
- [9] png_0.1-8 systemfonts_1.1.0
- [11] vctrs_0.6.5 reshape2_1.4.4
- [13] stringr_1.5.1 pkgconfig_2.0.3
- [15] SpatialExperiment_1.14.0 crayon_1.5.3
- [17] fastmap_1.2.0 backports_1.5.0
- [19] magick_2.8.4 XVector_0.44.0
- [21] labeling_0.4.3 utf8_1.2.4
- [23] rmarkdown_2.27 UCSC.utils_1.0.0
- [25] ragg_1.3.2 purrr_1.0.2
- [27] xfun_0.46 beachmat_2.20.0
- [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
- [31] jsonlite_1.8.8 DelayedArray_0.30.1
- [33] BiocParallel_1.38.0 terra_1.7-78
- [35] irlba_2.3.5.1 parallel_4.4.1
- [37] R6_2.5.1 stringi_1.8.4
- [39] RColorBrewer_1.1-3 reticulate_1.38.0
- [41] parallelly_1.37.1 GenomicRanges_1.56.1
- [43] scattermore_1.2 Rcpp_1.0.13
- [45] bookdown_0.40 SummarizedExperiment_1.34.0
- [47] knitr_1.48 future.apply_1.11.2
- [49] R.utils_2.12.3 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
+ [7] matrixStats_1.3.0 compiler_4.4.1
+ [9] png_0.1-8 systemfonts_1.1.0
+ [11] vctrs_0.6.5 reshape2_1.4.4
+ [13] stringr_1.5.1 pkgconfig_2.0.3
+ [15] SpatialExperiment_1.14.0 crayon_1.5.3
+ [17] fastmap_1.2.0 backports_1.5.0
+ [19] magick_2.8.4 XVector_0.44.0
+ [21] labeling_0.4.3 utf8_1.2.4
+ [23] rmarkdown_2.27 UCSC.utils_1.0.0
+ [25] ragg_1.3.2 purrr_1.0.2
+ [27] xfun_0.46 beachmat_2.20.0
+ [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
+ [31] jsonlite_1.8.8 DelayedArray_0.30.1
+ [33] BiocParallel_1.38.0 terra_1.7-78
+ [35] irlba_2.3.5.1 parallel_4.4.1
+ [37] R6_2.5.1 stringi_1.8.4
+ [39] RColorBrewer_1.1-3 reticulate_1.38.0
+ [41] parallelly_1.37.1 GenomicRanges_1.56.1
+ [43] scattermore_1.2 Rcpp_1.0.13
+ [45] bookdown_0.40 SummarizedExperiment_1.34.0
+ [47] knitr_1.48 future.apply_1.11.2
+ [49] R.utils_2.12.3 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/reference-keys.txt b/reference-keys.txt
index a17604d..555f956 100644
--- a/reference-keys.txt
+++ b/reference-keys.txt
@@ -4,115 +4,116 @@ fig:unnamed-chunk-14
fig:unnamed-chunk-15
fig:unnamed-chunk-16
fig:unnamed-chunk-18
-fig:unnamed-chunk-30
-fig:unnamed-chunk-35
-fig:unnamed-chunk-38
-fig:unnamed-chunk-40
+fig:unnamed-chunk-32
+fig:unnamed-chunk-37
fig:unnamed-chunk-42
-fig:unnamed-chunk-44
+fig:unnamed-chunk-45
+fig:unnamed-chunk-47
fig:unnamed-chunk-49
fig:unnamed-chunk-51
-fig:unnamed-chunk-53
-fig:unnamed-chunk-55
-fig:unnamed-chunk-59
-fig:unnamed-chunk-61
-fig:unnamed-chunk-63
+fig:unnamed-chunk-56
+fig:unnamed-chunk-58
+fig:unnamed-chunk-60
+fig:unnamed-chunk-62
fig:unnamed-chunk-66
-fig:unnamed-chunk-69
-fig:unnamed-chunk-74
+fig:unnamed-chunk-68
+fig:unnamed-chunk-70
+fig:unnamed-chunk-73
fig:unnamed-chunk-76
-fig:unnamed-chunk-78
-fig:unnamed-chunk-80
-fig:unnamed-chunk-82
-fig:unnamed-chunk-84
+fig:unnamed-chunk-81
+fig:unnamed-chunk-83
+fig:unnamed-chunk-85
fig:unnamed-chunk-87
+fig:unnamed-chunk-89
+fig:unnamed-chunk-91
fig:unnamed-chunk-94
-fig:unnamed-chunk-96
-fig:unnamed-chunk-98
fig:unnamed-chunk-101
fig:unnamed-chunk-103
fig:unnamed-chunk-105
+fig:unnamed-chunk-108
+fig:unnamed-chunk-110
fig:unnamed-chunk-112
-fig:unnamed-chunk-114
-fig:unnamed-chunk-118
-fig:unnamed-chunk-120
-fig:unnamed-chunk-123
-fig:unnamed-chunk-128
-fig:unnamed-chunk-131
-fig:unnamed-chunk-134
-fig:unnamed-chunk-137
+fig:unnamed-chunk-119
+fig:unnamed-chunk-121
+fig:unnamed-chunk-125
+fig:unnamed-chunk-127
+fig:unnamed-chunk-130
+fig:unnamed-chunk-135
+fig:unnamed-chunk-138
fig:unnamed-chunk-141
-fig:unnamed-chunk-143
-fig:unnamed-chunk-147
+fig:unnamed-chunk-144
+fig:unnamed-chunk-148
fig:unnamed-chunk-150
-fig:unnamed-chunk-152
+fig:unnamed-chunk-154
+fig:unnamed-chunk-157
fig:unnamed-chunk-159
-fig:unnamed-chunk-161
-fig:unnamed-chunk-163
-fig:unnamed-chunk-165
-fig:unnamed-chunk-167
-fig:unnamed-chunk-169
-fig:unnamed-chunk-171
+fig:unnamed-chunk-166
+fig:unnamed-chunk-168
+fig:unnamed-chunk-170
+fig:unnamed-chunk-172
fig:unnamed-chunk-174
-fig:unnamed-chunk-175
fig:unnamed-chunk-176
-fig:unnamed-chunk-185
-fig:unnamed-chunk-191
-fig:unnamed-chunk-194
+fig:unnamed-chunk-178
+fig:unnamed-chunk-181
+fig:unnamed-chunk-182
+fig:unnamed-chunk-183
+fig:unnamed-chunk-192
+fig:unnamed-chunk-198
fig:unnamed-chunk-201
-fig:unnamed-chunk-204
-fig:unnamed-chunk-206
-fig:unnamed-chunk-210
-fig:unnamed-chunk-212
-fig:unnamed-chunk-215
+fig:unnamed-chunk-208
+fig:unnamed-chunk-211
+fig:unnamed-chunk-213
fig:unnamed-chunk-217
fig:unnamed-chunk-219
-fig:unnamed-chunk-221
+fig:unnamed-chunk-222
fig:unnamed-chunk-224
-fig:unnamed-chunk-225
-fig:unnamed-chunk-227
-fig:unnamed-chunk-229
-fig:unnamed-chunk-230
+fig:unnamed-chunk-226
+fig:unnamed-chunk-228
+fig:unnamed-chunk-231
fig:unnamed-chunk-232
fig:unnamed-chunk-234
fig:unnamed-chunk-236
-fig:unnamed-chunk-275
-fig:unnamed-chunk-279
-fig:unnamed-chunk-281
-fig:unnamed-chunk-283
+fig:unnamed-chunk-237
+fig:unnamed-chunk-239
+fig:unnamed-chunk-241
+fig:unnamed-chunk-243
+fig:unnamed-chunk-282
fig:unnamed-chunk-286
fig:unnamed-chunk-288
fig:unnamed-chunk-290
-fig:unnamed-chunk-292
-fig:unnamed-chunk-294
+fig:unnamed-chunk-293
+fig:unnamed-chunk-295
fig:unnamed-chunk-297
fig:unnamed-chunk-299
fig:unnamed-chunk-301
-fig:unnamed-chunk-303
-fig:unnamed-chunk-305
-fig:unnamed-chunk-365
-fig:unnamed-chunk-366
+fig:unnamed-chunk-304
+fig:unnamed-chunk-306
+fig:unnamed-chunk-308
+fig:unnamed-chunk-310
+fig:unnamed-chunk-312
fig:unnamed-chunk-372
+fig:unnamed-chunk-373
fig:unnamed-chunk-379
-fig:unnamed-chunk-381
-fig:unnamed-chunk-387
-fig:unnamed-chunk-389
-fig:unnamed-chunk-395
-fig:unnamed-chunk-397
-fig:unnamed-chunk-462
-fig:unnamed-chunk-465
+fig:unnamed-chunk-386
+fig:unnamed-chunk-388
+fig:unnamed-chunk-394
+fig:unnamed-chunk-396
+fig:unnamed-chunk-402
+fig:unnamed-chunk-404
fig:unnamed-chunk-469
-fig:unnamed-chunk-470
fig:unnamed-chunk-472
-fig:unnamed-chunk-474
fig:unnamed-chunk-476
-fig:unnamed-chunk-478
+fig:unnamed-chunk-477
fig:unnamed-chunk-479
fig:unnamed-chunk-481
+fig:unnamed-chunk-483
fig:unnamed-chunk-485
+fig:unnamed-chunk-486
fig:unnamed-chunk-488
-fig:unnamed-chunk-490
fig:unnamed-chunk-492
+fig:unnamed-chunk-495
+fig:unnamed-chunk-497
+fig:unnamed-chunk-499
giotto-suite-workshop-2024
instructors
topics-and-schedule
@@ -153,6 +154,10 @@ presentation-1
ecosystem
installation-python-environment
giotto-installation-1
+system-prerequisites
+installation-of-released-version
+installation-of-dev-branch-giotto-packages
+common-install-issues
python-environment
default-installation
custom-installs
diff --git a/search_index.json b/search_index.json
index 2736898..72ab7c3 100644
--- a/search_index.json
+++ b/search_index.json
@@ -1 +1 @@
-[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: Ensure your system To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak. pak::pak("drieslab/Giotto") 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
+[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: 4.3.1.1 System prerequisites for windows, Rtools needs to be installed a major dependency terra needs GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux 4.3.1.2 Installation of released version To install the currently released version of Giotto in a single step: pak::pak("drieslab/Giotto") This should automatically install all the Giotto dependencies and other Giotto module packages. 4.3.1.3 Installation of dev branch Giotto packages You can install dev branch versions by using devtools::install_github() Core module dev branchs: “drieslab/Giotto@suite_dev” “drieslab/GiottoVisuals@dev” “drieslab/GiottoClass@dev” “drieslab/GiottoUtils@dev” devtools::install_github("drieslab/GiottoClass@dev") pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages. 4.3.1.4 Common install issues If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work. devtools::install_version("Matrix", version = "1.6-5") If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it. install.packages("irlba", type = "source") ## # Downloading packages ------------------------------------------------------- ## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s] ## Successfully downloaded 1 package in 1.2 seconds. ## ## The following package(s) will be installed: ## - irlba [2.3.5.1] ## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu". ## ## # Installing packages -------------------------------------------------------- ## - Installing irlba ... OK [built from source and cached in 5.7s] ## Successfully installed 1 package in 5.7 seconds. 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are: python_path - when set, will activate a python environment save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render. save_plot - whether to save plots to the save_dir return_plot - whether to return the plot objects. When FALSE, only NULL is returned show_plot - whether to show the plot in the viewer These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function. library(Giotto) save_dir <- "results/01_session2/" # this call will also intialize the python env instrs <- createGiottoInstructions( save_dir = save_dir, # working directory is the default show_plot = FALSE, save_plot = TRUE, return_plot = FALSE, python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default ) force(instrs) Giotto object creation functions all have an instructions param for passing in instructions objects. giotto objects will also respond to the instructions() generic. test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object # example plot g <- GiottoData::loadGiottoMini("visium") instructions(g) <- instrs instructions(g, "show_plot") # instructions say not to plot to viewer spatPlot2D(g, show_image = TRUE, image_name = "image") # instead it will directly write to the results folder As an example, you can also set individual instructions instructions(g, "show_plot") <- TRUE spatPlot2D(g, show_image = TRUE, image_name = "image") Figure 4.2: example image output "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
diff --git a/spatial-multi-modal-analysis.html b/spatial-multi-modal-analysis.html
index 5bdab74..c78f952 100644
--- a/spatial-multi-modal-analysis.html
+++ b/spatial-multi-modal-analysis.html
@@ -563,48 +563,48 @@ 13.2.3 Spatial classes:# load in data
-library(Giotto)
-
-g <- GiottoData::loadGiottoMini("vizgen")
-
-activeSpatUnit(g) <- "aggregate"
-
-gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE)
-
-gimg <- getGiottoImage(g)
+
13.3 Examples of the simple transforms with a giottoPolygon
-rain <- rainbow(nrow(gpoly))
-
-line_width <- 0.3
-
-# par to setup the grid plotting layout
-p <- par(no.readonly = TRUE)
-par(mfrow=c(3,3))
-gpoly |>
- plot(main = "no transform", col = rain, lwd = line_width)
-
-gpoly |> spatShift(dx = 1000) |>
- plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
-
-gpoly |> spin(45) |>
- plot(main = "spin(45)", col = rain, lwd = line_width)
-
-gpoly |> rescale(fx = 10, fy = 6) |>
- plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
-
-gpoly |> flip(direction = "vertical") |>
- plot(main = "flip()", col = rain, lwd = line_width)
-
-gpoly |> t() |>
- plot(main = "t()", col = rain, lwd = line_width)
-
-gpoly |> shear(fx = 0.5) |>
- plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
-par(p)
+rain <- rainbow(nrow(gpoly))
+
+line_width <- 0.3
+
+# par to setup the grid plotting layout
+p <- par(no.readonly = TRUE)
+par(mfrow=c(3,3))
+gpoly |>
+ plot(main = "no transform", col = rain, lwd = line_width)
+
+gpoly |> spatShift(dx = 1000) |>
+ plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
+
+gpoly |> spin(45) |>
+ plot(main = "spin(45)", col = rain, lwd = line_width)
+
+gpoly |> rescale(fx = 10, fy = 6) |>
+ plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
+
+gpoly |> flip(direction = "vertical") |>
+ plot(main = "flip()", col = rain, lwd = line_width)
+
+gpoly |> t() |>
+ plot(main = "t()", col = rain, lwd = line_width)
+
+gpoly |> shear(fx = 0.5) |>
+ plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
+par(p)
@@ -617,20 +617,20 @@ 13.4 Affine transforms# create affine2d
-aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
+# create affine2d
+aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set.
-
-# append several simple transforms
-aff <- aff |>
- spatShift(dx = 1000) |>
- spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- flip(direction = "vertical") |>
- t() |>
- shear(fx = 0.5)
-force(aff)
+
+# append several simple transforms
+aff <- aff |>
+ spatShift(dx = 1000) |>
+ spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ flip(direction = "vertical") |>
+ t() |>
+ shear(fx = 0.5)
+force(aff)
<affine2d>
anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax)
rotate : -0.785398163397448 (rad)
@@ -639,35 +639,35 @@ 13.4 Affine transformsplot(aff)
## RNA modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)
-
-## Protein modality
-visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "protein",
- type = "kNN",
- dimensions_to_use = 1:10,
- k = 20)## RNA modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)
+
+## Protein modality
+visium_tonsil <- createNearestNetwork(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "protein",
+ type = "kNN",
+ dimensions_to_use = 1:10,
+ k = 20)visium_tonsil <- runWNN(visium_tonsil,
- spat_unit = "cell",
- modality_1 = "rna",
- modality_2 = "protein",
- pca_name_modality_1 = "pca",
- pca_name_modality_2 = "protein.pca",
- k = 20,
- integrated_feat_type = NULL,
- matrix_result_name = NULL,
- w_name_modality_1 = NULL,
- w_name_modality_2 = NULL,
- verbose = TRUE)visium_tonsil <- runWNN(visium_tonsil,
+ spat_unit = "cell",
+ modality_1 = "rna",
+ modality_2 = "protein",
+ pca_name_modality_1 = "pca",
+ pca_name_modality_2 = "protein.pca",
+ k = 20,
+ integrated_feat_type = NULL,
+ matrix_result_name = NULL,
+ w_name_modality_1 = NULL,
+ w_name_modality_2 = NULL,
+ verbose = TRUE)visium_tonsil <- runIntegratedUMAP(visium_tonsil,
- modality1 = "rna",
- modality2 = "protein",
- spread = 7,
- min_dist = 1,
- force = FALSE)visium_tonsil <- runIntegratedUMAP(visium_tonsil,
+ modality1 = "rna",
+ modality2 = "protein",
+ spread = 7,
+ min_dist = 1,
+ force = FALSE)visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- nn_network_to_use = "kNN",
- network_name = "integrated_kNN",
- name = "integrated_leiden_clus",
- resolution = 1)visium_tonsil <- doLeidenCluster(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ nn_network_to_use = "kNN",
+ network_name = "integrated_kNN",
+ name = "integrated_leiden_clus",
+ resolution = 1)plotUMAP(gobject = visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- dim_reduction_name = "integrated.umap",
- point_size = 2,
- title = "Integrated UMAP using Integrated Leiden clusters")plotUMAP(gobject = visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ dim_reduction_name = "integrated.umap",
+ point_size = 2,
+ title = "Integrated UMAP using Integrated Leiden clusters")Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. @@ -850,14 +850,14 @@
14.8 Multi-omics integrationVisualize spatial plot with integrated clusters
Plot the spatial distribution of the integrated Leiden clusters.
-spatPlot2D(visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- point_size = 3,
- show_image = TRUE,
- title = "Integrated Leiden clustering")
-
+spatPlot2D(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ point_size = 3,
+ show_image = TRUE,
+ title = "Integrated Leiden clustering")
+
Figure 14.9: Spatial distribution of the integrated Leiden clusters.
@@ -866,85 +866,85 @@
14.8 Multi-omics integration
14.9 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
- [7] matrixStats_1.3.0 compiler_4.4.1
- [9] png_0.1-8 systemfonts_1.1.0
- [11] vctrs_0.6.5 reshape2_1.4.4
- [13] stringr_1.5.1 pkgconfig_2.0.3
- [15] SpatialExperiment_1.14.0 crayon_1.5.3
- [17] fastmap_1.2.0 backports_1.5.0
- [19] magick_2.8.4 XVector_0.44.0
- [21] labeling_0.4.3 utf8_1.2.4
- [23] rmarkdown_2.27 UCSC.utils_1.0.0
- [25] ragg_1.3.2 purrr_1.0.2
- [27] xfun_0.46 beachmat_2.20.0
- [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
- [31] jsonlite_1.8.8 DelayedArray_0.30.1
- [33] BiocParallel_1.38.0 terra_1.7-78
- [35] irlba_2.3.5.1 parallel_4.4.1
- [37] R6_2.5.1 stringi_1.8.4
- [39] RColorBrewer_1.1-3 reticulate_1.38.0
- [41] parallelly_1.37.1 GenomicRanges_1.56.1
- [43] scattermore_1.2 Rcpp_1.0.13
- [45] bookdown_0.40 SummarizedExperiment_1.34.0
- [47] knitr_1.48 future.apply_1.11.2
- [49] R.utils_2.12.3 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
+ [7] matrixStats_1.3.0 compiler_4.4.1
+ [9] png_0.1-8 systemfonts_1.1.0
+ [11] vctrs_0.6.5 reshape2_1.4.4
+ [13] stringr_1.5.1 pkgconfig_2.0.3
+ [15] SpatialExperiment_1.14.0 crayon_1.5.3
+ [17] fastmap_1.2.0 backports_1.5.0
+ [19] magick_2.8.4 XVector_0.44.0
+ [21] labeling_0.4.3 utf8_1.2.4
+ [23] rmarkdown_2.27 UCSC.utils_1.0.0
+ [25] ragg_1.3.2 purrr_1.0.2
+ [27] xfun_0.46 beachmat_2.20.0
+ [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
+ [31] jsonlite_1.8.8 DelayedArray_0.30.1
+ [33] BiocParallel_1.38.0 terra_1.7-78
+ [35] irlba_2.3.5.1 parallel_4.4.1
+ [37] R6_2.5.1 stringi_1.8.4
+ [39] RColorBrewer_1.1-3 reticulate_1.38.0
+ [41] parallelly_1.37.1 GenomicRanges_1.56.1
+ [43] scattermore_1.2 Rcpp_1.0.13
+ [45] bookdown_0.40 SummarizedExperiment_1.34.0
+ [47] knitr_1.48 future.apply_1.11.2
+ [49] R.utils_2.12.3 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/reference-keys.txt b/reference-keys.txt
index a17604d..555f956 100644
--- a/reference-keys.txt
+++ b/reference-keys.txt
@@ -4,115 +4,116 @@ fig:unnamed-chunk-14
fig:unnamed-chunk-15
fig:unnamed-chunk-16
fig:unnamed-chunk-18
-fig:unnamed-chunk-30
-fig:unnamed-chunk-35
-fig:unnamed-chunk-38
-fig:unnamed-chunk-40
+fig:unnamed-chunk-32
+fig:unnamed-chunk-37
fig:unnamed-chunk-42
-fig:unnamed-chunk-44
+fig:unnamed-chunk-45
+fig:unnamed-chunk-47
fig:unnamed-chunk-49
fig:unnamed-chunk-51
-fig:unnamed-chunk-53
-fig:unnamed-chunk-55
-fig:unnamed-chunk-59
-fig:unnamed-chunk-61
-fig:unnamed-chunk-63
+fig:unnamed-chunk-56
+fig:unnamed-chunk-58
+fig:unnamed-chunk-60
+fig:unnamed-chunk-62
fig:unnamed-chunk-66
-fig:unnamed-chunk-69
-fig:unnamed-chunk-74
+fig:unnamed-chunk-68
+fig:unnamed-chunk-70
+fig:unnamed-chunk-73
fig:unnamed-chunk-76
-fig:unnamed-chunk-78
-fig:unnamed-chunk-80
-fig:unnamed-chunk-82
-fig:unnamed-chunk-84
+fig:unnamed-chunk-81
+fig:unnamed-chunk-83
+fig:unnamed-chunk-85
fig:unnamed-chunk-87
+fig:unnamed-chunk-89
+fig:unnamed-chunk-91
fig:unnamed-chunk-94
-fig:unnamed-chunk-96
-fig:unnamed-chunk-98
fig:unnamed-chunk-101
fig:unnamed-chunk-103
fig:unnamed-chunk-105
+fig:unnamed-chunk-108
+fig:unnamed-chunk-110
fig:unnamed-chunk-112
-fig:unnamed-chunk-114
-fig:unnamed-chunk-118
-fig:unnamed-chunk-120
-fig:unnamed-chunk-123
-fig:unnamed-chunk-128
-fig:unnamed-chunk-131
-fig:unnamed-chunk-134
-fig:unnamed-chunk-137
+fig:unnamed-chunk-119
+fig:unnamed-chunk-121
+fig:unnamed-chunk-125
+fig:unnamed-chunk-127
+fig:unnamed-chunk-130
+fig:unnamed-chunk-135
+fig:unnamed-chunk-138
fig:unnamed-chunk-141
-fig:unnamed-chunk-143
-fig:unnamed-chunk-147
+fig:unnamed-chunk-144
+fig:unnamed-chunk-148
fig:unnamed-chunk-150
-fig:unnamed-chunk-152
+fig:unnamed-chunk-154
+fig:unnamed-chunk-157
fig:unnamed-chunk-159
-fig:unnamed-chunk-161
-fig:unnamed-chunk-163
-fig:unnamed-chunk-165
-fig:unnamed-chunk-167
-fig:unnamed-chunk-169
-fig:unnamed-chunk-171
+fig:unnamed-chunk-166
+fig:unnamed-chunk-168
+fig:unnamed-chunk-170
+fig:unnamed-chunk-172
fig:unnamed-chunk-174
-fig:unnamed-chunk-175
fig:unnamed-chunk-176
-fig:unnamed-chunk-185
-fig:unnamed-chunk-191
-fig:unnamed-chunk-194
+fig:unnamed-chunk-178
+fig:unnamed-chunk-181
+fig:unnamed-chunk-182
+fig:unnamed-chunk-183
+fig:unnamed-chunk-192
+fig:unnamed-chunk-198
fig:unnamed-chunk-201
-fig:unnamed-chunk-204
-fig:unnamed-chunk-206
-fig:unnamed-chunk-210
-fig:unnamed-chunk-212
-fig:unnamed-chunk-215
+fig:unnamed-chunk-208
+fig:unnamed-chunk-211
+fig:unnamed-chunk-213
fig:unnamed-chunk-217
fig:unnamed-chunk-219
-fig:unnamed-chunk-221
+fig:unnamed-chunk-222
fig:unnamed-chunk-224
-fig:unnamed-chunk-225
-fig:unnamed-chunk-227
-fig:unnamed-chunk-229
-fig:unnamed-chunk-230
+fig:unnamed-chunk-226
+fig:unnamed-chunk-228
+fig:unnamed-chunk-231
fig:unnamed-chunk-232
fig:unnamed-chunk-234
fig:unnamed-chunk-236
-fig:unnamed-chunk-275
-fig:unnamed-chunk-279
-fig:unnamed-chunk-281
-fig:unnamed-chunk-283
+fig:unnamed-chunk-237
+fig:unnamed-chunk-239
+fig:unnamed-chunk-241
+fig:unnamed-chunk-243
+fig:unnamed-chunk-282
fig:unnamed-chunk-286
fig:unnamed-chunk-288
fig:unnamed-chunk-290
-fig:unnamed-chunk-292
-fig:unnamed-chunk-294
+fig:unnamed-chunk-293
+fig:unnamed-chunk-295
fig:unnamed-chunk-297
fig:unnamed-chunk-299
fig:unnamed-chunk-301
-fig:unnamed-chunk-303
-fig:unnamed-chunk-305
-fig:unnamed-chunk-365
-fig:unnamed-chunk-366
+fig:unnamed-chunk-304
+fig:unnamed-chunk-306
+fig:unnamed-chunk-308
+fig:unnamed-chunk-310
+fig:unnamed-chunk-312
fig:unnamed-chunk-372
+fig:unnamed-chunk-373
fig:unnamed-chunk-379
-fig:unnamed-chunk-381
-fig:unnamed-chunk-387
-fig:unnamed-chunk-389
-fig:unnamed-chunk-395
-fig:unnamed-chunk-397
-fig:unnamed-chunk-462
-fig:unnamed-chunk-465
+fig:unnamed-chunk-386
+fig:unnamed-chunk-388
+fig:unnamed-chunk-394
+fig:unnamed-chunk-396
+fig:unnamed-chunk-402
+fig:unnamed-chunk-404
fig:unnamed-chunk-469
-fig:unnamed-chunk-470
fig:unnamed-chunk-472
-fig:unnamed-chunk-474
fig:unnamed-chunk-476
-fig:unnamed-chunk-478
+fig:unnamed-chunk-477
fig:unnamed-chunk-479
fig:unnamed-chunk-481
+fig:unnamed-chunk-483
fig:unnamed-chunk-485
+fig:unnamed-chunk-486
fig:unnamed-chunk-488
-fig:unnamed-chunk-490
fig:unnamed-chunk-492
+fig:unnamed-chunk-495
+fig:unnamed-chunk-497
+fig:unnamed-chunk-499
giotto-suite-workshop-2024
instructors
topics-and-schedule
@@ -153,6 +154,10 @@ presentation-1
ecosystem
installation-python-environment
giotto-installation-1
+system-prerequisites
+installation-of-released-version
+installation-of-dev-branch-giotto-packages
+common-install-issues
python-environment
default-installation
custom-installs
diff --git a/search_index.json b/search_index.json
index 2736898..72ab7c3 100644
--- a/search_index.json
+++ b/search_index.json
@@ -1 +1 @@
-[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: Ensure your system To install the currently released version of Giotto on R 4.4 in a single step, we recommend using pak. pak::pak("drieslab/Giotto") 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
+[["index.html", "Workshop: Spatial multi-omics data analysis with Giotto Suite 1 Giotto Suite Workshop 2024 1.1 Instructors 1.2 Topics and Schedule: 1.3 License", " Workshop: Spatial multi-omics data analysis with Giotto Suite Ruben Dries, Jiaji George Chen, Joselyn Cristina Chávez-Fuentes, Junxiang Xu ,Edward Ruiz, Jeff Sheridan, Iqra Amin, Wen Wang 1 Giotto Suite Workshop 2024 Workshop: Spatial multi-omics data analysis with Giotto Suite Github repo: https://github.com/drieslab/giotto_workshop_2024/ Giotto Suite Website: http://www.giottosuite.com Twitter/X: https://x.com/GiottoSpatial Code repo: https://github.com/drieslab/Giotto Issues page: https://github.com/drieslab/Giotto/issues Discussions page: https://github.com/drieslab/Giotto/discussions 1.1 Instructors Ruben Dries: Assistant Professor of Medicine at Boston University Joselyn Cristina Chávez Fuentes: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai Jiaji George Chen: Ph.D. student at Boston University Junxiang Xu: Ph.D. student at Boston University Edward C. Ruiz: Ph.D. student at Boston University Jeff Sheridan: Postdoctoral fellow at Boston University Iqra Amin: Bioinformatician at Boston University Wen Wang: Postdoctoral fellow at Icahn School of Medicine at Mount Sinai 1.2 Topics and Schedule: Day 1: Introduction Spatial omics technologies Spatial sequencing Spatial in situ Spatial proteomics spatial other: ATAC-seq, lipidomics, etc Introduction to the Giotto package Ecosystem Installation + python environment Giotto instructions Data formatting and Pre-processing Creating a Giotto object From matrix + locations From subcellular raw data (transcripts or images) + polygons Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) Spatial plots Subsetting: Based on IDs Based on locations Visualizations Introduction to spatial multi-modal dataset (10X Genomics breast cancer) and goal for the next days Quality control Statistics Normalization Feature selection: Highly Variable Features: loess regression binned pearson residuals Spatial variable genes Dimension Reduction PCA UMAP/t-SNE Visualizations Clustering Non-spatial k-means Hierarchical clustering Leiden/Louvain Spatial Spatial variable genes Spatial co-expression modules Day 2: Spatial Data Analysis Spatial sequencing based technology: Visium Differential expression Enrichment & Deconvolution PAGE/Rank SpatialDWLS Visualizations Interactive tools Spatial expression patterns Spatial variable genes Spatial co-expression modules Spatial HMRF Spatial sequencing based technology: Visium HD Tiling and aggregation Scalability (duckdb) and projection functions Spatial expression patterns Spatial co-expression module Spatial in situ technology: Xenium Read in raw data Transcript coordinates Polygon coordinates Visualizations Overlap txs & polygons Typical aggregated workflow Feature/molecule specific analysis Visualizations Transcript enrichment GSEA Spatial location analysis Spatial cell type co-localization analysis Spatial niche analysis Spatial niche trajectory analysis Visualizations Spatial proteomics: multiplex IF Read in raw data Intensity data (IF or any other image) Polygon coordinates Visualizations Overlap intensity & workflows Typical aggregated workflow Visualizations Day 3: Advanced Tutorials Multiple samples Create individual giotto objects Join Giotto Objects Perform Harmony and default workflows Visualizations Spatial multi-modal Co-registration of datasets Examples in giotto suite manuscript Multi-omics integration Example in giotto suite manuscript Interoperability w/ other frameworks AnnData/SpatialData SpatialExperiment Seurat Interoperability w/ isolated tools Spatial niche trajectory analysis Interactivity with the R/Spatial ecosystem Kriging Contributing to Giotto 1.3 License This material has a Creative Commons Attribution-ShareAlike 4.0 International License. To get more information about this license, visit http://creativecommons.org/licenses/by-sa/4.0/ "],["datasets-packages.html", "2 Datasets & Packages 2.1 Datasets to download 2.2 Needed packages", " 2 Datasets & Packages 2.1 Datasets to download Here we provide links to the original datasets that were used for this workshop. Some of the datasets were modified (e.g. downsampled or subsetted) for the purpose of this workshop. You can download them from their original source or download all of them - including intermediate files - from the following Zenodo repository: 2.1.1 Zenodo repository https://zenodo.org/communities/gw2024/records?q=&l=list&p=1&s=10&sort=newest 2.1.2 10X Genomics Visium Mouse Brain Section (Coronal) dataset https://support.10xgenomics.com/spatial-gene-expression/datasets/1.1.0/V1_Adult_Mouse_Brain 2.1.3 10X Genomics Visium HD: FFPE Human Colon Cancer https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-human-crc 2.1.4 10X Genomics multi-modal dataset https://www.10xgenomics.com/products/xenium-in-situ/preview-dataset-human-breast 2.1.5 10X Genomics multi-omics Visium CytAssist Human Tonsil dataset https://www.10xgenomics.com/resources/datasets/gene-protein-expression-library-of-human-tonsil-cytassist-ffpe-2-standard 2.1.6 10X Genomics Human Prostate Cancer Adenocarcinoma with Invasive Carcinoma (FFPE) https://www.10xgenomics.com/datasets/human-prostate-cancer-adenocarcinoma-with-invasive-carcinoma-ffpe-1-standard-1-3-0 2.1.7 10X Genomics Normal Human Prostate (FFPE) https://www.10xgenomics.com/datasets/normal-human-prostate-ffpe-1-standard-1-3-0 2.1.8 Xenium https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard 2.1.9 MERFISH cortex dataset https://doi.brainimagelibrary.org/doi/10.35077/g.21 2.2 Needed packages To run all the tutorials from this Giotto Suite workshop you will need to install additional R and Python packages. Here we provide detailed instructions and discuss some common difficulties with installing these packages. The easiest way would be to copy each code snippet into your R/Rstudio Console using fresh a R session. 2.2.1 CRAN dependencies: cran_dependencies <- c("BiocManager", "devtools", "pak") install.packages(cran_dependencies, Ncpus = 4) 2.2.2 terra installation terra may have some additional steps when installing depending on which system you are on. Please see the terra repo for specifics. Installations of the CRAN release on Windows and Mac are expected to be simple, only requiring the code below. For Linux, there are several prerequisite installs: GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 On our AlmaLinux 8 HPC, the following versions have been working well: gdal/3.6.4 geos/3.11.1 proj/9.2.0 sqlite3/3.37.2 install.packages("terra") 2.2.3 Matrix installation !! FOR R VERSIONS LOWER THAN 4.4.0 !! Giotto requires Matrix 1.6-2 or greater, but when installing Giotto with pak on an R version lower than 4.4.0, the installation can fail asking for R 4.5 which doesn’t exist yet. We can solve this by installing the 1.6-5 version directly by un-commenting and running the line below. # devtools::install_version("Matrix", version = "1.6-5") 2.2.4 Rtools installation Before installing Giotto on a windows PC please make sure to install the relevant version of Rtools. If you have a Mac or linux PC, or have already installed Rtools, please ignore this step. 2.2.5 Giotto installation pak::pak("drieslab/Giotto") pak::pak("drieslab/GiottoData") 2.2.6 irlba install Reinstall irlba from source. Avoids the common function 'as_cholmod_sparse' not provided by package 'Matrix' error. See this issue for more info. install.packages("irlba", type = "source") 2.2.7 arrow install arrow is a suggested package that we use here to open parquet files. The parquet files that 10X provides use zstd compression which the default arrow installation may not provide. has_arrow <- requireNamespace("arrow", quietly = TRUE) zstd <- TRUE if (has_arrow) { zstd <- arrow::arrow_info()$capabilities[["zstd"]] } if (!has_arrow || !zstd) { Sys.setenv(ARROW_WITH_ZSTD = "ON") install.packages("assertthat", "bit64") install.packages("arrow", repos = c("https://apache.r-universe.dev")) } 2.2.8 Bioconductor dependencies: bioc_dependencies <- c( "scran", "ComplexHeatmap", "SpatialExperiment", "ggspavis", "scater", "nnSVG" ) 2.2.9 CRAN packages: needed_packages_cran <- c( "dplyr", "gstat", "hdf5r", "miniUI", "shiny", "xml2", "future", "future.apply", "exactextractr", "tidyr", "viridis", "quadprog", "Rfast", "pheatmap", "patchwork", "Seurat", "harmony", "scatterpie", "R.utils", "qs" ) pak::pkg_install(c(bioc_dependencies, needed_packages_cran)) 2.2.10 Packages from GitHub github_packages <- c( "satijalab/seurat-data" ) pak::pkg_install(github_packages) 2.2.11 Python environments # default giotto environment Giotto::installGiottoEnvironment() reticulate::py_install( pip = TRUE, envname = 'giotto_env', packages = c( "scanpy" ) ) # install another environment with py 3.8 for cellpose reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) "],["spatial-omics-technologies.html", "3 Spatial omics technologies 3.1 Presentation 3.2 Short summary", " 3 Spatial omics technologies Ruben Dries August 5th 2024 3.1 Presentation 3.2 Short summary 3.2.1 Why do we need spatial omics technologies? Spatial omics allows us to examine the role of one or more cells within its normal context. This spatial context is typically organized at multiple length scales, and considers both adjacent neighboring cells and larger levels of tissue organization. Figure 3.1: Capturing tissue complexity with RNA-seq, scRNAseq, and Spatial Omics 3.2.2 What is spatial omics? Spatial omics is typically a combination of spatial sequencing and/or imaging together with understanding the obtained results through spatial data science. Figure 3.2: Spatial Omics Constituents 3.2.3 What are the main spatial omics technologies? The large majority - and most popular or accessible - spatial technologies are: - spatial antibody-multiplex proteomics - spatial multiplex in situ hybridization (ISH)-based transcriptomics - spatial sequencing-based transcriptomics Figure 3.3: Lewis et al. Nat Meth Review. Characteristics of spatial omics technologies 3.2.4 Other Spatial omics: ATAC-seq, CUT&Tag, lipidomics, etc A growing number of other spatial technologies exist that profile different types of molecular analytes. One example is using a deterministic barcoding approach (Rong Fan’s group) to explore open (ATAC-seq) or modified (CUT&Tag) chromatin in a spatially aware manner. Figure 3.4: Vandereyken et al. Nat Rev Genetics. Spatial deterministic barcoding for ATAC-seq and CUT&tag 3.2.5 What are the different types of spatial downstream analyses? There exist a large and diverse amount of different downstream spatial data analyses that use different available data types and formats as input. Figure 3.5: Dries, R. et al. Genome Res. Downstream analysis in spatial data analysis. "],["introduction-to-the-giotto-package.html", "4 Introduction to the Giotto package 4.1 Presentation 4.2 Ecosystem 4.3 Installation + python environment 4.4 Giotto instructions", " 4 Introduction to the Giotto package Ruben Dries & Jiaji George Chen August 5th 2024 4.1 Presentation 4.2 Ecosystem Giotto Suite is a modular ecosystem of individual R packages that each provide different functionality and that together provide users with a fully integrated spatial multi-omics workflow. Figure 4.1: Overview of the modular Giotto Suite ecosystem Each package also has its own website: - GiottoUtils: https://drieslab.github.io/GiottoUtils/ - GiottoClass: https://drieslab.github.io/GiottoClass/ - GiottoData: https://drieslab.github.io/GiottoData/ - GiottoVisuals: https://drieslab.github.io/GiottoVisuals/ More information is available at https://drieslab.github.io/Giotto_website/articles/ecosystem.html 4.3 Installation + python environment 4.3.1 Giotto installation Giotto Suite is currently available installable only from GitHub, but we are actively working on getting it into a major repository. Much of this already covered in Section 2.2, but the highlights are: 4.3.1.1 System prerequisites for windows, Rtools needs to be installed a major dependency terra needs GDAL (>= 2.2.3), GEOS (>= 3.4.0), PROJ (>= 4.9.3), sqlite3 on linux 4.3.1.2 Installation of released version To install the currently released version of Giotto in a single step: pak::pak("drieslab/Giotto") This should automatically install all the Giotto dependencies and other Giotto module packages. 4.3.1.3 Installation of dev branch Giotto packages You can install dev branch versions by using devtools::install_github() Core module dev branchs: “drieslab/Giotto@suite_dev” “drieslab/GiottoVisuals@dev” “drieslab/GiottoClass@dev” “drieslab/GiottoUtils@dev” devtools::install_github("drieslab/GiottoClass@dev") pak tends to forcibly install all dependencies, which can have issues when working with multiple dev branch packages. 4.3.1.4 Common install issues If installing on an R version earlier than 4.4, pak can throw errors when installing Matrix. To get around this, install Matrix v1.6-5 and then installing Giotto with pak should work. devtools::install_version("Matrix", version = "1.6-5") If you come across the function 'as_cholmod_sparse' not provided by package 'Matrix' error when running Giotto, reinstalling irlba from source may resolve it. install.packages("irlba", type = "source") ## # Downloading packages ------------------------------------------------------- ## - Downloading irlba from CRAN ... OK [228.2 Kb in 0.36s] ## Successfully downloaded 1 package in 1.2 seconds. ## ## The following package(s) will be installed: ## - irlba [2.3.5.1] ## These packages will be installed into "~/work/giotto_workshop_2024/giotto_workshop_2024/renv/library/linux-ubuntu-jammy/R-4.4/x86_64-pc-linux-gnu". ## ## # Installing packages -------------------------------------------------------- ## - Installing irlba ... OK [built from source and cached in 5.7s] ## Successfully installed 1 package in 5.7 seconds. 4.3.2 Python environment 4.3.2.1 Default installation In order to make use of python packages, the first thing to do after installing Giotto for the first time is to create a giotto python environment. Giotto provides the following as a convenience wrapper around reticulate functions to setup a default environment. library(Giotto) installGiottoEnvironment() Two things are needed for python to work: A conda (e.g. miniconda or anaconda) installation which is the package and environment management system. Independent environment(s) with specific versions of the python language and associated python packages. installGiottoEnvironment() checks both and will install miniconda using reticulate if necessary. If a specific conda binary already exists that you want to use, the conda param can be set, or you can set the reticulate option options(\"reticulate.conda_binary\" = \"[conda path]\") or Sys.setenv(\"RETICULATE_CONDA\" = \"[conda path]\"). After ensuring the conda binary exists, the default Giotto environment is installed which is a python 3.10.2 environment named ‘giotto_env’. It will contain several default packages that Giotto installs: “pandas==1.5.1” “networkx==2.8.8” “python-igraph==0.10.2” “leidenalg==0.9.0” “python-louvain==0.16” “python.app==1.4” (if needed) “scikit-learn==1.1.3” 4.3.2.2 Custom installs Custom python environments can be made by first setting up a new environment and establishing the name and python version to use. reticulate::conda_create(envname = "[name of env]", python_version = ???) Following that, one or more python packages to install can be added to the environment. reticulate::py_install( pip = TRUE, envname = '[name of env]', packages = c( "package1", "package2", "..." ) ) Once an environment has been set up, Giotto can hook into it. 4.3.2.3 Using a specific environment When using python through reticulate, R only allows one environment to be activated per session. Once a session has loaded a python environment, it can no longer switch to another one. Giotto activates a python environment when any of the following happens: a giotto object is created giottoInstructions are created (createGiottoInstructions()) GiottoClass::set_giotto_python_path() is called (most straightforward) Which environment is activated is based on a set of 5 defaults in decreasing priority. User provided (when python_path param is given. Either a full filepath or an env name are accepted.) Any provided path or envname set in options options(\"giotto.py_path\" = \"[path to env or envname]\") Default expected giotto environment location based on reticulate::miniconda_path() Envname \"giotto_env\" System default python environment Method 2 is most recommended when there is a non-standard python environment to regularly use with Giotto. You would run file.edit(\"~/.Rprofile\") and then add options(\"giotto.py_path\" = \"[path to env or envname]\") as a line so that it is automatically set at the start of each session. If a specific environment should only be used a couple times then method 1 is easiest: GiottoClass::set_giotto_python_path(python_path = "[path to env or envname]") To check which conda environments exist on your machine: reticulate::conda_list() Once an environment is activated, you can check more details and ensure that it is the one you are expecting by running: reticulate::py_config() 4.4 Giotto instructions Giotto uses giottoInstructions in order to set a behavior for a particular giotto object. Most commonly used are: python_path - when set, will activate a python environment save_dir - save directory to use. Usually for plots generated. This can help speed things up since the viewer no longer has to render. save_plot - whether to save plots to the save_dir return_plot - whether to return the plot objects. When FALSE, only NULL is returned show_plot - whether to show the plot in the viewer These objects are created with createGiottoInstructions() and the created objects can be edited afterwards using the instructions() generic function. library(Giotto) save_dir <- "results/01_session2/" # this call will also intialize the python env instrs <- createGiottoInstructions( save_dir = save_dir, # working directory is the default show_plot = FALSE, save_plot = TRUE, return_plot = FALSE, python_path = NULL # when NULL, this calls GiottoClass::set_giotto_python_path() to get the default ) force(instrs) Giotto object creation functions all have an instructions param for passing in instructions objects. giotto objects will also respond to the instructions() generic. test <- giotto(instructions = instrs) # passing NULL instead will also generate a default instructions object # example plot g <- GiottoData::loadGiottoMini("visium") instructions(g) <- instrs instructions(g, "show_plot") # instructions say not to plot to viewer spatPlot2D(g, show_image = TRUE, image_name = "image") # instead it will directly write to the results folder As an example, you can also set individual instructions instructions(g, "show_plot") <- TRUE spatPlot2D(g, show_image = TRUE, image_name = "image") Figure 4.2: example image output "],["data-formatting-and-pre-processing.html", "5 Data formatting and Pre-processing 5.1 Data formats 5.2 Pre-processing", " 5 Data formatting and Pre-processing Jiaji George Chen August 5th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session3" 5.1 Data formats .h5 .mtx - get10xmatrix .parquet .csv/.tsv - fread .json -jsonlite .geojson .tiff/.ome.tif 5.2 Pre-processing necessary additional packages? "],["creating-a-giotto-object.html", "6 Creating a Giotto object 6.1 Overview 6.2 From matrix + locations 6.3 From subcellular raw data (transcripts or images) + polygons 6.4 From piece-wise 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) 6.6 Spatial plots 6.7 Subsetting 6.8 Mini objects & GiottoData", " 6 Creating a Giotto object Jiaji George Chen August 5th 2024 6.0.1 Used packages Giotto GiottoClass GiottoData pak save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/01_session4" 6.1 Overview Giotto has representations for both aggregated (cell by count) + xy(z) information and subcellular polygon or mask information and transcript xy(z) or image information. A Giotto object can be set up from either of the two above sets of information. 6.2 From matrix + locations createGiottoObject() 6.3 From subcellular raw data (transcripts or images) + polygons createGiottoObjectSubcellular() The above two methods accept data, converts them into Giotto’s compatible formats, and then assembles the final giotto object. If desired you can also assemble a giotto object piece-wise after creating the Giotto subobjects manually. 6.4 From piece-wise g <- giotto() g <- setGiotto(g, ??) 6.5 Using convenience functions for popular technologies (Vizgen, Xenium, CosMx, …) There are also several convenience functions we provide for loading in data from popular platforms. Many of these will be touched on later during other sessions createGiottoVisiumObject() createGiottoVisiumHDObject() createGiottoXeniumObject() createGiottoCosMxObject() createGiottoMerscopeObject() 6.6 Spatial plots Giotto has several spatial plotting functions. At the lowest level, you directly call plot() on several subobjects in order to see what they look like, particularly the ones containing spatial info. gpoints <- GiottoData::loadSubObjectMini("giottoPoints") gpoly <- GiottoData::loadSubObjectMini("giottoPolygon") spatlocs <- GiottoData::loadSubObjectMini("spatLocsObj") spatnet <- GiottoData::loadSubObjectMini("spatialNetworkObj") pca <- GiottoData::loadSubObjectMini("dimObj") plot(pca, dims = c(3,10)) 6.7 Subsetting Based on IDs Based on locations Visualizations 6.8 Mini objects & GiottoData Giotto makes available several mini objects to allow devs and users to work with easily loadable Giotto objects. These are small subsets of a larger dataset that often contain some worked through analyses and are fully functional. pak::pak("drieslab/GiottoData") "],["visium-part-i.html", "7 Visium Part I 7.1 The Visium technology 7.2 Introduction to the spatial dataset 7.3 Download dataset 7.4 Create the Giotto object 7.5 Subset on spots that were covered by tissue 7.6 Quality control 7.7 Filtering 7.8 Normalization 7.9 Feature selection 7.10 Dimension Reduction 7.11 Clustering 7.12 Save the object 7.13 Session info", " 7 Visium Part I Joselyn Cristina Chávez Fuentes August 5th 2024 7.1 The Visium technology Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression. Figure 7.1: Visum workflow. Source: 10X Genomics You can use standard fixation and staining techniques, including hematoxylin and eosin (H&E) staining, to visualize tissue sections on slides using a brightfield microscope and immunofluorescence (IF) staining to visualize protein detection in tissue sections on slides using a fluorescent microscope. 7.2 Introduction to the spatial dataset The visium fresh frozen mouse brain tissue (Strain C57BL/6) dataset was obtained from 10X genomics. The tissue was embedded and cryosectioned as described in Visium Spatial Protocols - Tissue Preparation Guide (Demonstrated Protocol CG000240). Tissue sections of 10 µm thickness from a slice of the coronal plane were placed on Visium Gene Expression Slides. You can find more information about his sample here 7.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/01_session5/") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”. untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", exdir = "data/01_session5") untar(tarfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz", exdir = "data/01_session5") 7.4 Create the Giotto object createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object. library(Giotto) ## Set instructions results_folder <- "results/01_session5/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## Provide the path to the visium folder data_path <- "data/01_session5/" ## Create object directly from the visium folder visium_brain <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) 7.5 Subset on spots that were covered by tissue Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area. spatPlot2D( gobject = visium_brain, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE) Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0). Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area. metadata <- getCellMetadata(gobject = visium_brain, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_brain <- subsetGiotto(gobject = visium_brain, cell_ids = in_tissue_barcodes) 7.6 Quality control Statistics Use the function addStatistics() to count the number of features per spot. The statistics information will be stored in the metadata table under the new column “nr_feats”. Then, use this column to visualize the number of features per spot across the sample. visium_brain_statistics <- addStatistics(gobject = visium_brain, expression_values = "raw") ## visualize spatPlot2D(gobject = visium_brain_statistics, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.3: Spatial distribution of features per spot. filterDistributions() creates a histogram to show the distribution of features per spot across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "cells") Figure 7.4: Distribution of features per spot. When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample. filterDistributions(gobject = visium_brain_statistics, detection = "feats") Figure 7.5: Distribution of cells with different features per spot. filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data: filterCombinations(gobject = visium_brain_statistics, expression_thresholds = c(1, 2, 3), feat_det_in_min_cells = c(50, 100, 200), min_det_feats_per_cell = c(500, 1000, 1500)) Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. 7.7 Filtering Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample. visium_brain <- filterGiotto( gobject = visium_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE ) Feature type: rna Number of cells removed: 4 out of 2702 Number of feats removed: 7311 out of 22125 7.8 Normalization Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one. visium_brain <- normalizeGiotto( gobject = visium_brain, scalefactor = 6000, verbose = TRUE ) Calculate the normalized number of features per spot and save the statistics in the metadata table. visium_brain <- addStatistics(gobject = visium_brain) ## visualize spatPlot2D(gobject = visium_brain, cell_color = "nr_feats", color_as_factor = FALSE) Figure 7.7: Spatial distribution of the number of features per spot. 7.9 Feature selection 7.9.1 Highly Variable Features: Calculating Highly Variable Features (HVF) is necessary to identify genes (or features) that display significant variability across the spots. There are a few methods to choose from depending on the underlying distribution of the data: loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_loess", save_plot = TRUE, default_save_name = "HVFplot_loess") Figure 7.8: Covariance of HVFs using the loess method. pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes. visium_brain <- calculateHVF(gobject = visium_brain, method = "var_p_resid", save_plot = TRUE, default_save_name = "HVFplot_pearson") Figure 7.9: Variance of HVFs using the pearson residuals method. binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function. visium_brain <- calculateHVF(gobject = visium_brain, method = "cov_groups", save_plot = TRUE, default_save_name = "HVFplot_binned") Figure 7.10: Covariance of HVFs using the binned method. 7.10 Dimension Reduction 7.10.1 PCA Principal Components Analysis (PCA) is applied to reduce the dimensionality of gene expression data by transforming it into principal components, which are linear combinations of genes ranked by the variance they explain, with the first components capturing the most variance. runPCA() will look for the previous calculation of highly variable features, stored as a column in the feature metadata. If the HVF labels are not found in the giotto object, then runPCA() will use all the features available in the sample to calculate the Principal Components. visium_brain <- runPCA(gobject = visium_brain) You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument. my_features <- head(getFeatureMetadata(visium_brain, output = "data.table")$feat_ID, 1000) visium_brain <- runPCA(gobject = visium_brain, feats_to_use = my_features, name = "custom_pca") Visualization Create a screeplot to visualize the percentage of variance explained by each component. screePlot(gobject = visium_brain, ncp = 30) Figure 7.11: Screeplot showing the variance explained per principal component. Visualized the PCA calculated using the HVFs. plotPCA(gobject = visium_brain) Figure 7.12: PCA plot using HVFs. Visualized the custom PCA calculated using the vector of features. plotPCA(gobject = visium_brain, dim_reduction_name = "custom_pca") Figure 7.13: PCA using custom features. Unlike PCA, Uniform Manifold Approximation and Projection (UMAP) and t-Stochastic Neighbor Embedding (t-SNE) do not assume linearity. After running PCA, UMAP or t-SNE allows you to visualize the dataset in 2D. 7.10.2 UMAP visium_brain <- runUMAP(visium_brain, dimensions_to_use = 1:10) Visualization plotUMAP(gobject = visium_brain) Figure 7.14: UMAP using the 10 first principal components. 7.10.3 t-SNE visium_brain <- runtSNE(gobject = visium_brain, dimensions_to_use = 1:10) Visualization plotTSNE(gobject = visium_brain) Figure 7.15: tSNE using the 10 first principal components. 7.11 Clustering Create a sNN network (default) visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15) Create a kNN network visium_brain <- createNearestNetwork(gobject = visium_brain, dimensions_to_use = 1:10, k = 15, type = "kNN") 7.11.1 Calculate Leiden clustering Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters. visium_brain <- doLeidenCluster(gobject = visium_brain, resolution = 0.4, n_iterations = 1000) Visualization plotPCA(gobject = visium_brain, cell_color = "leiden_clus") Figure 7.16: PCA plot, colors indicate the Leiden clusters. Use the cluster IDs to visualize the clusters in the UMAP space. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = FALSE, point_size = 2.5) Figure 7.17: UMAP plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotUMAP(gobject = visium_brain, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2.5) Figure 7.18: UMAP showing the nearest network. Use the cluster IDs to visualize the clusters on the tSNE. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5) Figure 7.19: tSNE plot, colors indicate the Leiden clusters. Set the argument “show_NN_network = TRUE” to visualize the connections between spots. plotTSNE(gobject = visium_brain, cell_color = "leiden_clus", point_size = 2.5, show_NN_network = TRUE) Figure 7.20: tSNE showing the nearest network. Use the cluster IDs to visualize their spatial location. spatPlot2D(visium_brain, cell_color = "leiden_clus", point_size = 3) Figure 7.21: Spatial plot, colors indicate the Leiden clusters. 7.11.2 Calculate Louvain clustering Louvain is an alternative clustering method, used to detect communities in large networks. visium_brain <- doLouvainCluster(visium_brain) spatPlot2D(visium_brain, cell_color = "louvain_clus") Figure 7.22: Spatial plot, colors indicate the Louvain clusters. You can find more information about the differences between the Leiden and Louvain methods in this paper: From Louvain to Leiden: guaranteeing well-connected communities, 2019 7.12 Save the object saveGiotto(visium_brain, "visium_brain_object") 7.13 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] GiottoUtils_0.1.10 matrixStats_1.3.0 [7] compiler_4.4.1 png_0.1-8 [9] systemfonts_1.1.0 vctrs_0.6.5 [11] reshape2_1.4.4 stringr_1.5.1 [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0 [15] crayon_1.5.3 fastmap_1.2.0 [17] backports_1.5.0 magick_2.8.4 [19] XVector_0.44.0 labeling_0.4.3 [21] utf8_1.2.4 rmarkdown_2.27 [23] UCSC.utils_1.0.0 ragg_1.3.2 [25] purrr_1.0.2 xfun_0.46 [27] beachmat_2.20.0 zlibbioc_1.50.0 [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8 [31] DelayedArray_0.30.1 BiocParallel_1.38.0 [33] terra_1.7-78 irlba_2.3.5.1 [35] parallel_4.4.1 R6_2.5.1 [37] stringi_1.8.4 RColorBrewer_1.1-3 [39] reticulate_1.38.0 parallelly_1.37.1 [41] GenomicRanges_1.56.1 scattermore_1.2 [43] Rcpp_1.0.13 bookdown_0.40 [45] SummarizedExperiment_1.34.0 knitr_1.48 [47] future.apply_1.11.2 R.utils_2.12.3 [49] FNN_1.1.4 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["visium-part-ii.html", "8 Visium Part II 8.1 Load the object 8.2 Differential expression 8.3 Enrichment & Deconvolution 8.4 Spatial expression patterns 8.5 Spatially informed clusters 8.6 Spatial domains HMRF 8.7 Interactive tools 8.8 Session info", " 8 Visium Part II Joselyn Cristina Chávez Fuentes August 6th 2024 8.1 Load the object library(Giotto) visium_brain <- loadGiotto("visium_brain_object") 8.2 Differential expression 8.2.1 Gini markers The Gini method identifies genes that are very selectively expressed in a specific cluster, however not always expressed in all cells of that cluster. In other words, highly specific but not necessarily sensitive at the single-cell level. Calculate the top marker genes per cluster using the gini method. gini_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "gini", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_gini), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.1: Violin plot showing the top gini genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_gini), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_gini)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 15, base_height = 20)) Figure 8.3: Spatial distribution of the top gini genes scaled expression. 8.2.2 Scran markers The Scran method is preferred for robust differential expression analysis, especially when addressing technical variability or differences in sequencing depth across spatial locations. [redo] Calculate the top marker genes per cluster using the scran method scran_markers <- findMarkers_one_vs_all(gobject = visium_brain, method = "scran", expression_values = "normalized", cluster_column = "leiden_clus", min_feats = 10) topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats Visualize Plot the normalized expression distribution of the top expressed genes. violinPlot(visium_brain, feats = unique(topgenes_scran), cluster_column = "leiden_clus", strip_text = 6, strip_position = "right", save_param = list(base_width = 5, base_height = 30)) Figure 8.4: Violin plot of the top scran genes normalized expression. Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster. plotMetaDataHeatmap(visium_brain, selected_feats = unique(topgenes_scran), metadata_cols = "leiden_clus", x_text_size = 10, y_text_size = 10) Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster. Visualize the scaled expression spatial distribution of the top expressed genes across the sample. dimFeatPlot2D(visium_brain, expression_values = "scaled", feats = sort(unique(topgenes_scran)), cow_n_col = 5, point_size = 1, save_param = list(base_width = 20, base_height = 20)) Figure 8.6: Spatial distribution of the top scran genes scaled expression. In practice, it is often beneficial to apply both Gini and Scran methods and compare results for a more complete understanding of differential gene expression across clusters. 8.3 Enrichment & Deconvolution Visium spatial transcriptomics does not provide single-cell resolution, making cell type annotation a harder problem. Giotto provides several ways to calculate enrichment of specific cell-type signature gene lists. Download the single-cell dataset GiottoData::getSpatialDataset(dataset = "scRNA_mouse_brain", directory = "data/02_session1/") Create the single-cell object and run the normalization step results_folder <- "results/02_session1/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, python_path = python_path ) sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz" sc_metadata <- "data/02_session1/brain_sc_metadata.csv" giotto_SC <- createGiottoObject(expression = sc_expression, instructions = instructions) giotto_SC <- addCellMetadata(giotto_SC, new_metadata = data.table::fread(sc_metadata)) giotto_SC <- normalizeGiotto(giotto_SC) 8.3.1 PAGE/Rank Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset. Calculate the cell type markers markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC, method = "scran", expression_values = "normalized", cluster_column = "Class", min_feats = 3) top_markers <- markers_scran[, head(.SD, 10), by = "cluster"] celltypes <- levels(factor(markers_scran$cluster)) Create the signature matrix sign_list <- list() for (i in 1:length(celltypes)){ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats } sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes, sign_list = sign_list) Run the enrichment test with PAGE visium_brain <- runPAGEEnrich(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters. cell_types_PAGE <- colnames(sign_matrix) plotMetaDataCellsHeatmap(gobject = visium_brain, metadata_cols = "leiden_clus", value_cols = cell_types_PAGE, spat_enr_names = "PAGE", x_text_size = 8, y_text_size = 8) Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method. Plot the spatial distribution of the cell types. spatCellPlot2D(gobject = visium_brain, spat_enr_names = "PAGE", cell_annotation_values = cell_types_PAGE, cow_n_col = 3, coord_fix_ratio = 1, point_size = 1, show_legend = TRUE) Figure 8.8: Spatial distribution of cell types identified using the PAGE method. 8.3.2 SpatialDWLS Spatial Dampened Weighted Least Squares (DWLS) estimates the proportions of different cell types across spots in a tissue. Create the signature matrix sign_matrix <- makeSignMatrixDWLSfromMatrix( matrix = getExpression(giotto_SC, values = "normalized", output = "matrix"), cell_type = pDataDT(giotto_SC)$Class, sign_gene = top_markers$feats) Run the DWLS Deconvolution This step may take a couple of minutes to run. visium_brain <- runDWLSDeconv(gobject = visium_brain, sign_matrix = sign_matrix) Visualize Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot. spatDeconvPlot(visium_brain, show_image = FALSE, radius = 50, save_param = list(save_name = "8_spat_DWLS_pie_plot")) Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. 8.4 Spatial expression patterns 8.4.1 Spatial variable genes Create a spatial network visium_brain <- createSpatialNetwork(gobject = visium_brain, method = "kNN", k = 6, maximum_distance_knn = 400, name = "spatial_network") spatPlot2D(gobject = visium_brain, show_network= TRUE, network_color = "blue", spatial_network_name = "spatial_network") Figure 8.10: Spatial network across spots in the Visium mouse sample. Rank binarization Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not. This step may take a few minutes to run. ranktest <- binSpect(visium_brain, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "spatial_network") Visualize top results Plot the scaled expression of genes with the highest probability of being spatial genes. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = ranktest$feats[1:6], cow_n_col = 2, point_size = 1) Figure 8.11: Spatial distribution of the top spatial genes scaled expression. 8.4.2 Spatial co-expression modules Cluster the top 500 spatial genes into 20 clusters ext_spatial_genes <- ranktest[1:500,]$feats Use detectSpatialCorGenes function to calculate pairwise distances between genes. spat_cor_netw_DT <- detectSpatialCorFeats( visium_brain, method = "network", spatial_network_name = "spatial_network", subset_feats = ext_spatial_genes) Identify most similar spatially correlated genes for one gene top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Mbp", show_top_feats = 10) Visualize Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp. spatFeatPlot2D(visium_brain, expression_values = "scaled", feats = top10_genes$variable[1:4], point_size = 1.5) Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. Cluster spatial genes spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 20) Visualize clusters Plot the correlation of the top 500 spatial genes with their assigned cluster. heatmSpatialCorFeats(visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL)) Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. Rank spatial correlated clusters and show genes for selected clusters netw_ranks <- rankSpatialCorGroups( visium_brain, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus") Plot the correlation and number of spatial genes in each cluster. top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", selected_clusters = 6, show_top_feats = 1) Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. Create the metagene enrichment score per co-expression cluster cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT, use_clus_name = "spat_netw_clus", show_top_feats = 1) cluster_genes <- cluster_genes_DT$clus names(cluster_genes) <- cluster_genes_DT$feat_ID visium_brain <- createMetafeats(visium_brain, feat_clusters = cluster_genes, name = "cluster_metagene") Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster. spatCellPlot(visium_brain, spat_enr_names = "cluster_metagene", cell_annotation_values = netw_ranks$clusters, point_size = 1, cow_n_col = 5) Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. 8.5 Spatially informed clusters Get the top 30 genes per spatial co-expression cluster coexpr_dt <- data.table::data.table( genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus), cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus) data.table::setorder(coexpr_dt, cluster) top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster] spatial_genes <- top30_coexpr_dt$genes Re-calculate the clustering Use the spatial genes to calculate again the principal components, umap, network and clustering visium_brain <- runPCA(gobject = visium_brain, feats_to_use = spatial_genes, name = "custom_pca") visium_brain <- runUMAP(visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, name = "custom_umap") visium_brain <- createNearestNetwork(gobject = visium_brain, dim_reduction_name = "custom_pca", dimensions_to_use = 1:20, k = 5, name = "custom_NN") visium_brain <- doLeidenCluster(gobject = visium_brain, network_name = "custom_NN", resolution = 0.15, n_iterations = 1000, name = "custom_leiden") Visualize Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes. spatPlot2D(visium_brain, cell_color = "custom_leiden", point_size = 3) Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes. Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes. plotUMAP(gobject = visium_brain, cell_color = "custom_leiden") Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. 8.6 Spatial domains HMRF Hidden Markov Random Field (HMRF) models capture spatial dependencies and segment tissue regions based on shared and gene expression patterns. Do HMRF with different betas on top 30 genes per spatial co-expression module This step may take several minutes to run. HMRF_spatial_genes <- doHMRF(gobject = visium_brain, expression_values = "scaled", spatial_genes = spatial_genes, k = 20, spatial_network_name = "spatial_network", betas = c(0, 10, 5), output_folder = "11_HMRF/") Add the HMRF results to the giotto object visium_brain <- addHMRF(gobject = visium_brain, HMRFoutput = HMRF_spatial_genes, k = 20, betas_to_add = c(0, 10, 20, 30, 40), hmrf_name = "HMRF") Visualize Plot the spatial distribution of the HMRF domains. spatPlot2D(gobject = visium_brain, cell_color = "HMRF_k20_b.40") Figure 8.18: Spatial distribution of HMRF domains. 8.7 Interactive tools We have integrated a shiny app in Giotto to interactively select regions of a spatial plot. Create a spatial plot brain_spatPlot <- spatPlot2D(gobject = visium_brain, cell_color = "leiden_clus", show_image = FALSE, return_plot = TRUE, point_size = 1) brain_spatPlot Run the Shiny app plotInteractivePolygons(brain_spatPlot) Figure 8.19: Shiny app using the visium brain sample. Select the regions of interest and save the coordinates polygon_coordinates <- plotInteractivePolygons(brain_spatPlot) Figure 8.20: Polygons selected using the interactive Shiny app. Transform the data.table or data.frame with coordinates into a Giotto polygon object giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates, name = "selections", calc_centroids = TRUE) Add the polygons to the Giotto object visium_brain <- addGiottoPolygons(gobject = visium_brain, gpolygons = list(giotto_polygons)) Add the corresponding polygon IDs to the cell metadata visium_brain <- addPolygonCells(visium_brain, polygon_name = "selections") Extract the coordinates and IDs from cells located within one or multiple regions of interest. getCellsFromPolygon(visium_brain, polygon_name = "selections", polygons = "polygon 1") If no polygon name is provided, the function will retrieve cells located within all polygons getCellsFromPolygon(visium_brain, polygon_name = "selections") Compare the expression levels of some genes of interest between the selected regions comparePolygonExpression(visium_brain, selected_feats = c("Stmn1", "Psd", "Ly6h")) Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. Calculate the top genes expressed within each region, then provide the result to compare polygons scran_results <- findMarkers_one_vs_all( visium_brain, spat_unit = "cell", feat_type = "rna", method = "scran", expression_values = "normalized", cluster_column = "selections", min_feats = 2) top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats comparePolygonExpression(visium_brain, selected_feats = top_genes) Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. Compare the abundance of cell types between the selected regions compareCellAbundance(visium_brain) Figure 8.23: Heatmap showing the cell abundance per selected polygon. Use other columns within the cell metadata table to compare the cell type abundances compareCellAbundance(visium_brain, cell_type_column = "custom_leiden") Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. Use the spatPlot arguments to isolate and plot each region. spatPlot2D(visium_brain, cell_color = "leiden_clus", group_by = "selections", cow_n_col = 3, point_size = 2, show_legend = FALSE) Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. Color each cell by cluster, cell type or expression level. spatFeatPlot2D(visium_brain, expression_values = "scaled", group_by = "selections", feats = "Psd", point_size = 2) Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. Plot again the polygons plotPolygons(visium_brain, polygon_name = "selections", x = brain_spatPlot) Figure 8.27: Spatial location of selected polygons. 8.8 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] later_1.3.2 tibble_3.2.1 [3] R.oo_1.26.0 polyclip_1.10-7 [5] lifecycle_1.0.4 edgeR_4.2.1 [7] doParallel_1.0.17 lattice_0.22-6 [9] MASS_7.3-61 backports_1.5.0 [11] magrittr_2.0.3 sass_0.4.9 [13] limma_3.60.4 plotly_4.10.4 [15] rmarkdown_2.27 jquerylib_0.1.4 [17] yaml_2.3.9 metapod_1.12.0 [19] httpuv_1.6.15 sp_2.1-4 [21] reticulate_1.38.0 cowplot_1.1.3 [23] RColorBrewer_1.1-3 abind_1.4-5 [25] zlibbioc_1.50.0 quadprog_1.5-8 [27] GenomicRanges_1.56.1 purrr_1.0.2 [29] R.utils_2.12.3 BiocGenerics_0.50.0 [31] tweenr_2.0.3 circlize_0.4.16 [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1 [35] S4Vectors_0.42.1 ggrepel_0.9.5 [37] irlba_2.3.5.1 terra_1.7-78 [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0 [41] colorRamp2_0.1.0 codetools_0.2-20 [43] DelayedArray_0.30.1 scuttle_1.14.0 [45] ggforce_0.4.2 tidyselect_1.2.1 [47] shape_1.4.6.1 UCSC.utils_1.0.0 [49] farver_2.1.2 ScaledMatrix_1.12.0 [51] matrixStats_1.3.0 stats4_4.4.1 [53] GiottoData_0.2.12.0 jsonlite_1.8.8 [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0 [57] progressr_0.14.0 iterators_1.0.14 [59] systemfonts_1.1.0 foreach_1.5.2 [61] dbscan_1.2-0 tools_4.4.1 [63] ragg_1.3.2 Rcpp_1.0.13 [65] glue_1.7.0 SparseArray_1.4.8 [67] xfun_0.46 MatrixGenerics_1.16.0 [69] GenomeInfoDb_1.40.1 dplyr_1.1.4 [71] withr_3.0.0 fastmap_1.2.0 [73] bluster_1.14.0 fansi_1.0.6 [75] digest_0.6.36 rsvd_1.0.5 [77] R6_2.5.1 mime_0.12 [79] textshaping_0.4.0 colorspace_2.1-0 [81] scattermore_1.2 Cairo_1.6-2 [83] gtools_3.9.5 R.methodsS3_1.8.2 [85] utf8_1.2.4 tidyr_1.3.1 [87] generics_0.1.3 data.table_1.15.4 [89] FNN_1.1.4 httr_1.4.7 [91] htmlwidgets_1.6.4 S4Arrays_1.4.1 [93] scatterpie_0.2.3 uwot_0.2.2 [95] pkgconfig_2.0.3 gtable_0.3.5 [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4 [99] SingleCellExperiment_1.26.0 XVector_0.44.0 [101] htmltools_0.5.8.1 bookdown_0.40 [103] clue_0.3-65 scales_1.3.0 [105] Biobase_2.64.0 GiottoUtils_0.1.10 [107] png_0.1-8 SpatialExperiment_1.14.0 [109] scran_1.32.0 ggfun_0.1.5 [111] knitr_1.48 rstudioapi_0.16.0 [113] reshape2_1.4.4 rjson_0.2.21 [115] checkmate_2.3.1 cachem_1.1.0 [117] GlobalOptions_0.1.2 stringr_1.5.1 [119] parallel_4.4.1 miniUI_0.1.1.1 [121] RcppZiggurat_0.1.6 pillar_1.9.0 [123] grid_4.4.1 vctrs_0.6.5 [125] promises_1.3.0 BiocSingular_1.20.0 [127] beachmat_2.20.0 xtable_1.8-4 [129] cluster_2.1.6 evaluate_0.24.0 [131] magick_2.8.4 cli_3.6.3 [133] locfit_1.5-9.10 compiler_4.4.1 [135] rlang_1.1.4 crayon_1.5.3 [137] labeling_0.4.3 plyr_1.8.9 [139] stringi_1.8.4 viridisLite_0.4.2 [141] deldir_2.0-4 BiocParallel_1.38.0 [143] munsell_0.5.1 lazyeval_0.2.2 [145] Matrix_1.7-0 sparseMatrixStats_1.16.0 [147] ggplot2_3.5.1 statmod_1.5.0 [149] SummarizedExperiment_1.34.0 Rfast_2.1.0 [151] memoise_2.0.1 igraph_2.0.3 [153] bslib_0.7.0 RcppParallel_5.1.8 "],["visium-hd.html", "9 Visium HD 9.1 Objective 9.2 Background 9.3 Data Ingestion 9.4 Tiling and aggregation 9.5 Scalability and projection functions 9.6 Spatial expression patterns 9.7 Spatial co-expression modules", " 9 Visium HD Edward C. Ruiz August 6th 2024 9.1 Objective This tutorial demonstrates how to process Visium HD data at the highest 2 micron bin resolution using Giotto Suite and methods from the [dbverse] (link). 9.2 Background 9.2.1 Visium HD Technology Figure 9.1: Overview of Visium HD. Source: 10X Genomics Visium HD is a spatial transcriptomics technology recently developed by 10X Genomics. Details about this platform are discussed on the official 10X Genomics Visium HD website and the preprint by Oliveira et al. 2024 on bioRxiv. At the highest resolution, Visium HD data contains a 2 micron bin size. At the lowest resolution, Visium HD data is binned at a 16 micron bin size after running the spaceranger pipeline. 9.2.2 Colorectal Cancer Sample Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics For this tutorial we will be using the publicly available Colorectal Cancer Visium HD dataset. Details about this dataset and a link to download the raw data can be found at the 10X Genomics website. 9.3 Data Ingestion 9.3.1 Visium HD output data format Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. Visium HD data processed with the spaceranger pipeline is organized in this format containing various files associated with the sample. The files highlighted in yellow are what we will be using to read in these datasets. 9.3.2 Read in raw data # get10Xmatrix() # GiottoData:: 9.3.3 Giotto Visium HD convenience function We have also developed a convenience function to read in Visium HD data directly into Giotto. This function will read in the data and create a Giotto Object. # importVisiumHD() 9.4 Tiling and aggregation The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells. # tessellate() 9.5 Scalability and projection functions filter and normalization workflow PCA projection 9.6 Spatial expression patterns plotting 9.7 Spatial co-expression modules binspect? "],["xenium-1.html", "10 Xenium 10.1 Introduction to spatial dataset 10.2 Data prep 10.3 Convenience function 10.4 Piecewise loading 10.5 Xenium Images 10.6 Spatial aggregation 10.7 Aggregate analyses workflow 10.8 Niche clustering 10.9 Cell proximity enrichment 10.10 Pseudovisium", " 10 Xenium Jiaji George Chen August 6th 2024 10.1 Introduction to spatial dataset This is the 10X Xenium FFPE Human Lung Cancer dataset. Xenium captures individual transcript detections with a spatial resolution of 100s of nanometers, providing an extremely highly resolved subcellular spatial dataset. This particular dataset also showcases their recent multimodal cell segmentation outputs. The Xenium Human Multi-Tissue and Cancer Panel (377) genes was used. The exported data is from their Xenium Onboard Analysis v2.0.0 pipeline. The full data for this example can be found here: here The relevant items are: Xenium Output Bundle (full) Supplemental: Post-Xenium H&E image (OME-TIFF) Supplemental: H&E Image Alignment File (CSV) Additional package requirements When working with this data and trying to open the parquet files, you will need arrow built with ZTSD support. See the datasets & packages section for specific install instructions. 10.1.1 Output directory structure ├── analysis.tar.gz ├── analysis.zarr.zip ├── analysis_summary.html ├── aux_outputs.tar.gz ├── transcripts.csv.gz ├── transcripts.parquet ├── transcripts.zarr.zip ├── cell_boundaries.csv.gz ├── cell_boundaries.parquet ├── nucleus_boundaries.csv.gz ├── nucleus_boundaries.parquet ├── cell_feature_matrix.tar.gz ├── cell_feature_matrix │ ├── barcodes.tsv.gz │ ├── features.tsv.gz │ └── matrix.mtx.gz ├── cell_feature_matrix.h5 ├── cell_feature_matrix.zarr.zip ├── cells.csv.gz ├── cells.parquet ├── cells.zarr.zip ├── experiment.xenium ├── gene_panel.json ├── metrics_summary.csv ├── morphology.ome.tif ├── morphology_focus │ ├── morphology_focus_0000.ome.tif │ ├── morphology_focus_0001.ome.tif │ ├── morphology_focus_0002.ome.tif │ ├── morphology_focus_0003.ome.tif ├── Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif └── Xenium_V1_humanLung_Cancer_FFPE_he_imagealignment.csv The above directory structuring and naming is characteristic of Xenium v2.0 pipeline outputs. The only items that may not be exactly the same across all outputs are the morphology focus directory and the naming of the aligned image items. For the morphology focus images, you may have fewer images if the experiment did not include the multimodal cell segmentation. As for the aligned images, this is usually done after the Xenium experiment concludes and is added on using Xenium Explorer. Naming and location of the aligned image (he_image.ome.tif) and associated alignment info he_imagealignment.csv are entirely up to the user. 10.1.2 Mini Xenium Dataset library(Giotto) # set up paths data_path <- "data/02_session3/" save_dir <- "results/02_session3/" dir.create(save_dir, recursive = TRUE) # download the mini dataset and untar options("timeout" = Inf) download.file( url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1", destfile = file.path(save_dir, "workshop_xenium.zip") ) untar(tarfile = file.path(save_dir, "workshop_xenium.zip"), exdir = data_path) In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset. - Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax) - Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax) Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset 10.2 Data prep 10.2.1 Image conversion (may change) First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package. The image files can then be loaded in downstream steps. These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below. # image_paths <- list.files( # data_path, pattern = "morphology_focus|he_image.ome", # recursive = TRUE, full.names = TRUE # ) ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory. Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has. # lapply(image_paths, function(img) { # GiottoClass::ometif_to_tif(img, overwrite = TRUE) # }) We are also working on a method of directly accessing the ome.tifs for better compatibility in the future. 10.3 Convenience function Giotto has flexible methods for working with the Xenium outputs. The createGiottoXeniumObject() will generate a giotto object in a single step when provided the output directory. The default behavior is to load: transcripts information cell and nucleus boundaries feature metadata (gene_panel.json) For the full dataset (HPC): time: 1-2min | memory: 24GBC ?createGiottoXeniumObject g <- createGiottoXeniumObject(xenium_dir = data_path) # set instructions instructions(g, "save_dir") <- save_dir instructions(g, "save_plot") <- TRUE There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them. 10.3.1 Specific filepaths expression_path = , cell_metadata_path = , transcript_path = , bounds_path = , gene_panel_json_path = , The convenience function auto-detects filepaths based on the Xenium directory path and the preferred file formats .parquet for tabular (vs .csv) .h5 for matrix over other formats when available (vs .mtx) .zarr is currently not supported. When you need to use a different file format or something is not in the expected output structure, you can supply a specific filepath to the convenience function using these params. 10.3.2 Quality value qv_threshold = 20 # default The Quality Value is a Phred-based 0-40 value that 10X provides for every detection in their transcripts output. Higher values mean higher confidence in the decoded transcript identity. By default 10X uses a cutoff of QV = 20 for transcripts to use downstream. _*setting a value other than 20 will make the loaded dataset different from the 10X-provided expression matrix and cell metadata._ QV Calculation Raw Q-score based on how likely it is that an observed code is to be the codeword that it gets mapped to vs less likely codeword. Adjustment of raw Q-score by binning the transcripts by Q-value then adjusting the exact Q per bin based on proportion of Negative Control Codewords detected within. further info 10.3.3 Transcript type splitting feat_type = c( "rna", "NegControlProbe", "UnassignedCodeword", "NegControlCodeword" ), split_keyword = list( c("NegControlProbe"), c("UnassignedCodeword"), c("NegControlCodeword)" ) There are 4 types of transcript detections that 10X reports with their v2.0 pipeline: Gene expression - This is the rna gene detections Negative Control Codeword - (QC) Codewords that do not map to genes, but are in the codebook. Used to determine specificity of decoding algorithm. Negative Control Probe - (QC) Probes in panel but target non-biological sequences. Used to determine specificity of assay. Unassigned Codeword - (QC) Codewords that should not be used in the current panel With V3 on their Xenium prime outputs, there is additionally: Genomic Control Codeword (QC) Probes for intergenic genomic DNA instead of transcripts The main thing to watch out for is that the other probe types should be separated out from the the Gene expression or rna feature type. How to deal with these different types of detections is easily adjustable. With the feat_type param you declare which categories/feat_types you want to split transcript detections into. Then with split_keyword, you provide a list of character vectors containing grep() terms to search for. Note that there are 4 feat_types declared in this set of defaults, but 3 items passed to split_keyword. Any transcripts not matched by items in split_keyword, get categorized as the first provided feat_type (“rna”). 10.3.4 Centroids calculation Several Giotto operations require that a set of centroids are calculated for polygon spatial units. g <- addSpatialCentroidLocations(g, poly_info = "cell") g <- addSpatialCentroidLocations(g, poly_info = "nucleus") 10.3.5 Simple visualization spatInSituPlotPoints(g, polygon_feat_type = "cell", feats = list(rna = head(featIDs(g))), use_overlap = FALSE, polygon_color = "cyan", polygon_line_size = 0.1 ) Figure 10.2: Simple subcellular plotting to check data 10.4 Piecewise loading Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility. x <- importXenium(data_path) force(x) Giotto <XeniumReader> dir : data/02_session3/ qv_cutoff : 20 filetype : transcripts -- parquet boundaries -- parquet expression -- h5 cell_meta -- parquet funs : load_transcripts() load_polys() load_cellmeta() load_featmeta() load_expression() load_image() load_aligned_image() create_gobject() 10.4.1 load giottoPoints transcripts x$qv <- 20 # default tx <- x$load_transcripts() plot(tx[[1]]$rna, dens = TRUE) Figure 10.3: plot of Gene expression (‘rna’) density force(tx[[1]]$rna) rm(tx) # save space An object of class giottoPoints feat_type : "rna" Feature Information: class : SpatVector geometry : points dimensions : 479097, 10 (geometries, attributes) extent : 6000.001, 7000, -2200, -1400.012 (xmin, xmax, ymin, ymax) coord. ref. : names : feat_ID transcript_id cell_id overlaps_nucleus z_location qv fov_name type : <chr> <chr> <chr> <int> <num> <num> <chr> values : FBLN1 281487861612869 mcnjadoe-1 0 19.32 40 B11 PDGFRB 281487861612872 mcnjbidl-1 1 18.75 40 B11 PDGFRB 281487861612873 mcnjbidl-1 1 18.74 40 B11 nucleus_distance codeword_index feat_ID_uniq <num> <int> <int> 0 334 1 0 289 2 0 289 3 10.4.2 (optional) Loading pre-aggregated data Giotto can spatially aggregate the transcripts information based on a provided set of boundaries information, however 10X also provides a pre-aggregated set of cell by feature information and metadata. These values may be slightly different from those calculated by Giotto’s pipeline, and are not loaded by default. Some care needs to be taken when loading this information: The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function. The qv_threshold used must be 20 since the 10X outputs are based on that cutoff. x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset ex <- x$load_expression() featType(ex) [1] "rna" "Negative Control Probe" "Negative Control Codeword" [4] "Unassigned Codeword" The feature types here do not match what we established for the transcripts, so we can just change them. force(g) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] spatial locations ---------------- [cell] raw [nucleus] raw featType(ex[[2]]) <- c("NegControlProbe") featType(ex[[3]]) <- c("NegControlCodeword") featType(ex[[4]]) <- c("UnassignedCodeword") Then we can just append them to the Giotto object. Here we set up a second object since we will be using Giotto’s own aggregation method downstream. g2 <- g # append the expression info g2 <- setGiotto(g2, ex) # load cell metadata cx <- x$load_cellmeta() g2 <- setGiotto(g2, cx) force(g2) An object of class giotto >Active spat_unit: cell >Active feat_type: rna [SUBCELLULAR INFO] polygons : cell nucleus features : rna NegControlProbe UnassignedCodeword NegControlCodeword [AGGREGATE INFO] expression ----------------------- [cell][rna] raw [cell][NegControlProbe] raw [cell][NegControlCodeword] raw [cell][UnassignedCodeword] raw spatial locations ---------------- [cell] raw [nucleus] raw spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "cell_area", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) spatInSituPlotPoints(g2, # polygon shading params polygon_fill = "transcript_counts", polygon_fill_as_factor = FALSE, polygon_fill_gradient_style = "sequential", # polygon line params polygon_color = "grey", polygon_line_size = 0.1 ) Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’ rm(g2) # save space 10.5 Xenium Images Xenium outputs have several image outputs. For this dataset: morphology.ome.tif is a z-stacked image of the DAPI staining, with z levels separated as pages within the ome.tif. In this dataset, only pages 6 and 7 are really in focus. morphology_focus is a folder containing single-channel image(s), but with the original z information collapsed into a single in-focus layer. For all datasets, image 0000 will be DAPI staining, but if you have additional stains, such as the multimodal segmentation, they will also be here. These are the recommended immunofluorescence staining images to import. Xenium_V1_humanLung_Cancer_FFPE_he_image.ome.tif is an added on (in this case H&E) image with manual affine registration. 10.5.1 Image metadata The morphology_focus directory may contain multiple images, but to know more information, we have to check the ome.tif xml metadata. With a normal dataset, you can use: `GiottoClass::ometif_metadata([filepath], node = "Channel")` on one of the morphology_focus images, but since the mini dataset images are pre-processed, there is only an exported .xml to explore. The output of the code chunk below is the same as that from calling ometif_metadata() and looking for the Channel node. img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml") omemeta <- xml2::read_xml(img_xml_path) res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta)) res <- Reduce(rbind, xml2::xml_attrs(res)) rownames(res) <- NULL res <- as.data.frame(res) force(res) ID Name SamplesPerPixel 1 Channel:0 DAPI 1 2 Channel:1 18S 1 3 Channel:2 ATP1A1/CD45/E-Cadherin 1 4 Channel:3 alphaSMA/Vimentin 1 10.5.2 Image loading morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset). Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E These transforms are normally done automatically when using: # convenience function params load_images = list( img1 = "[img_path1.tif]", img2 = "[img_path2.tif]", img3 = "..." ), load_aligned_images = list( aligned_img = c( "[path to image.tif]", "[path to magealignment.csv]" ) ) # importer params x$load_image(path = "[img_path1.tif]", name = "img1") x$load_image(path = "[img_path2.tif]", name = "img2") ... x$load_aligned_image( path = "[path to image.tif]", imagealignment_path = "[path to magealignment.csv]", name = "aligned_img" ) Specifically for the aligned image, there is also read10xAffineImage() which has similar params, but also asks for the micron scaling factor. But for the mini dataset, the images are pre-processed and can be directly added. img_paths <- c( sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3), "data/02_session3/he_mini.tif" ) img_list <- createGiottoLargeImageList( img_paths, # naming is based on the channel metadata above names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"), use_rast_ext = TRUE, verbose = FALSE ) # make some images brighter img_list[[1]]@max_window <- 5000 img_list[[2]]@max_window <- 5000 img_list[[3]]@max_window <- 5000 # append images to gobject g <- setGiotto(g, img_list) # example plots spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "DAPI", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "18S", polygon_feat_type = "cell", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) spatInSituPlotPoints(g, show_image = TRUE, image_name = "ATP1A1/CD45/E-Cadherin", polygon_feat_type = "nucleus", polygon_color = "cyan", polygon_line_size = 0.1, polygon_alpha = 0 ) Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right) 10.6 Spatial aggregation First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix(). g <- calculateOverlap(g, spatial_info = "cell", feat_info = "rna" ) g <- overlapToMatrix(g) 10.7 Aggregate analyses workflow 10.7.1 Filtering # very permissive filtering. Mainly for removing 0 values g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 5 ) Feature type: rna Number of cells removed: 0 out of 7129 Number of feats removed: 0 out of 377 10.7.2 Normalization g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, polygon_fill = "nr_feats", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) spatInSituPlotPoints(g, polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential", polygon_fill_as_factor = FALSE ) Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right) When there are a lot of features, we would also select only the interesting highly variable features so that downstream dimension reduction has more meaningful separation. Here we skip HVF detection since there are only 377 genes. 10.7.3 Dimension Reduction Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering. g <- runPCA(g, feats_to_use = NULL) # feats_to_use = NULL since there are no HVFs calculated. Use all genes. screePlot(g, ncp = 30) Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 40 # default ) plotPCA(g) plotUMAP(g) Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right) 10.7.4 Clustering g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 40 ) # takes roughly 1 min to run g <- doLeidenCluster(g) plotPCA_3D(g, cell_color = "leiden_clus", point_size = 1, ) plotUMAP(g, cell_color = "leiden_clus", point_size = 0.1, point_shape = "no_border" ) Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) Figure 10.11: Spatial plot with leiden clustering annotations. 10.8 Niche clustering Building on top of these leiden annotations, we can define spatial niche signatures based on which leiden types are often found together. 10.8.1 Spatial network First a spatial network must be generated so that spatial relationships between cells can be understood. g <- createSpatialNetwork(g, method = "Delaunay" ) spatPlot2D(g, point_shape = "no_border", show_network = TRUE, point_size = 0.1, point_alpha = 0.5, network_color = "grey" ) Figure 10.12: Delaunay spatial network` 10.8.2 Niche calculation Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche. The results are stored as a new spatial enrichment entry called ‘leiden_niche’ g <- calculateSpatCellMetadataProportions(g, spat_network = "Delaunay_network", metadata_column = "leiden_clus", name = "leiden_niche" ) 10.8.3 kmeans clustering based on niche signature # retrieve the niche info prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table") # convert to matrix prop_matrix <- GiottoUtils::dt_to_matrix(prop_table) # perform kmeans clustering set.seed(1234) # make kmeans clustering reproducible prop_kmeans <- kmeans( x = prop_matrix, centers = 7, # controls how many clusters will be formed iter.max = 1000, nstart = 100 ) prop_kmeansDT = data.table::data.table( cell_ID = names(prop_kmeans$cluster), niche = prop_kmeans$cluster ) # return kmeans clustering on niche to gobject g <- addCellMetadata(g, new_metadata = prop_kmeansDT, by_column = TRUE, column_cell_ID = "cell_ID" ) # visualize niches spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_fill = "niche", # polygon_fill_code = getColors("Accent", 8), polygon_alpha = 1, polygon_fill_as_factor = TRUE ) ggplot2::ggplot( cx, ggplot2::aes(fill = as.character(leiden_clus), y = 1, x = as.character(niche))) + ggplot2::geom_bar(position = "fill", stat = "identity") + ggplot2::scale_fill_manual(values = c( "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628", "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69") ) Figure 10.13: Leiden annotation-based spatial niches Figure 10.14: Stacked barplot of leiden annotation composition by niche 10.9 Cell proximity enrichment Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions. # uses a lot of memory leiden_prox <- cellProximityEnrichment(g, cluster_column = "leiden_clus", spatial_network_name = "Delaunay_network", adjust_method = "fdr", number_of_simulations = 2000 ) cellProximityBarplot(g, CPscore = leiden_prox, min_orig_ints = 5, # minimum original cell-cell interactions min_sim_ints = 5 # minimum simulated cell-cell interactions ) Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right) Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out. # set up colors other_cell_color <- rep("grey", 12) int_6_8 <- int_2_9 <- other_cell_color int_6_8[c(6, 8)] <- c("orange", "cornflowerblue") int_2_9[c(2, 9)] <- c("orange", "cornflowerblue") spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_6_8, polygon_line_size = 0.1, polygon_alpha = 1, show_image = TRUE, image_name = "HE" ) spatInSituPlotPoints(g, polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, polygon_fill_code = int_2_9, polygon_line_size = 0.1, show_image = TRUE, polygon_alpha = 1, image_name = "HE" ) Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions 10.10 Pseudovisium Another thing we can do is create a “pseudovisium” dataset by tessellating across this dataset using the same layout and resolution as a Visium capture array. makePseudoVisium generates a Visium array of circular polygons across the spatial extent provided. Here we use ext() with the prefer arg pointing to the polygon and points data and all_data = TRUE, meaning that the combined spatial extent of those two data types will be returned, giving a good measure of where all the data in the object is at the moment. micron_size = 1 since the Xenium data is already scaled to microns. pvis <- makePseudoVisium( extent = ext(g, prefer = c("polygon", "points"), all_data = TRUE # this is the default ), micron_size = 1 ) g <- setGiotto(g, pvis) g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium") plot(pvis) Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium() 10.10.1 Pseudovisium aggregation and workflow Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow activeSpatUnit(g) <- "pseudo_visium" g <- calculateOverlap(g, spatial_info = "pseudo_visium", feat_info = "rna" ) g <- overlapToMatrix(g) g <- filterGiotto(g, expression_threshold = 1, feat_det_in_min_cells = 1, min_det_feats_per_cell = 100 ) g <- normalizeGiotto(g) g <- addStatistics(g) spatInSituPlotPoints(g, show_image = TRUE, image_name = "HE", polygon_feat_type = "pseudo_visium", polygon_fill = "total_expr", polygon_fill_gradient_style = "sequential" ) Figure 10.19: Pseudo visium total detections per spot g <- runPCA(g, feats_to_use = NULL) g <- runUMAP(g, dimensions_to_use = seq(15), n_neighbors = 15 ) g <- createNearestNetwork(g, dimensions_to_use = seq(15), k = 15 ) g <- doLeidenCluster(g, resolution = 1.5) # plots plotPCA(g, cell_color = "leiden_clus", point_size = 2) plotUMAP(g, cell_color = "leiden_clus", point_size = 2) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE ) spatInSituPlotPoints(g, polygon_feat_type = "pseudo_visium", polygon_fill = "leiden_clus", polygon_fill_as_factor = TRUE, show_image = TRUE, image_name = "HE" ) Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right) "],["spatial-proteomics-multiplexed-immunofluorescence.html", "11 Spatial proteomics: Multiplexed Immunofluorescence 11.1 Spatial Proteomics Technologies 11.2 Raw data type coming out of different technologies 11.3 Cell Segmentation file to get single cell level protein expression 11.4 Create a Giotto Object using list of gitto large images and polygons", " 11 Spatial proteomics: Multiplexed Immunofluorescence Junxiang Xu August 6th 2024 Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so. #Install the Giotto Environment with Cellpose, note that we only need to do it once reticulate::conda_create(envname = "giotto_cellpose", python_version = 3.8) #.re.restartR() reticulate::use_condaenv('giotto_cellpose') reticulate::py_install( pip = TRUE, envname = 'giotto_cellpose', packages = c( "pandas", "networkx", "python-igraph", "leidenalg", "scikit-learn", "cellpose", "smfishhmrf", 'tifffile', 'scikit-image' ) ) #.rs.restartR() Now, activate the Giotto python environment. #.rs.restartR() # Activate the Giotto python environment of your choice GiottoClass::set_giotto_python_path('giotto_cellpose') # Check if cellpose was successfully installed GiottoUtils::package_check('cellpose',repository = 'pip') 11.1 Spatial Proteomics Technologies This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. This is the example Lunaphore dataset from Lunaphore the official website and we are using the cropped one small area as an example. This is an overview of a subset of how the data would look like. 11.2 Raw data type coming out of different technologies 11.2.1 Use ome.tiff as an example output data to begin with OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore. library(Giotto) instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'), save_plot = TRUE, show_plot = TRUE, python_path = 'giotto_cellpose') options(timeout = 9999999) download_dir =file.path(getwd(),'/data/02_session4/') destfile = file.path(download_dir,'Lunaphore.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir) list.files(paste0(download_dir,'/Lunaphore')) We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML. img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame") img_meta However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it. ##Get channel information Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff") xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml")) node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata)) channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node))) channel_df 11.2.2 Use single channel images as an example output data to begin with Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like. #Create multichannel raster and extract each single channels Luna_terra = terra::rast(Luna) names(Luna_terra) <- channel_df$Name gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T) plot(gimg_DAPI) Extract and save the raster image for future use. single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/') if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) } for (i in 1:nrow(channel_df)){ single_channel <- terra::subset(Luna_terra, i) terra::writeRaster(single_channel, filename = paste0(single_channel_dir,names(single_channel),'.tiff'), overwrite=T) } Create a list of GiottoLargeImages using single channel rasters. file_names = list.files(single_channel_dir,full.names = TRUE) image_names = sub("\\\\.tiff$", "", list.files(single_channel_dir)) gimg_list <- createGiottoLargeImageList(raster_objects = file_names, names = image_names, negative_y = F, flip_vertical = T) names(gimg_list) <- image_names plot(gimg_list[['Vimentin']]) 11.3 Cell Segmentation file to get single cell level protein expression Cell segmentation is necessary to generate single cell level protein expression. Currently, there are multiple algorithms to generate segmentations from images and output could be different. For that purpose, Giotto provides createGiottoPolygonsFromMask(), createGiottoPolygonsFromDfr(), createGiottoPolygonsFromGeoJSON() to load different type of file to the giottoPolygon Class. 11.3.1 Using segmentation output file from DeepCell(mesmer) as an example. We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function. gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T) plot(gpoly_mesmer) We can also zoom in to check how does the segmentation look. zoom <- c(2000,2500,2000,2500) plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) 11.3.2 Using Giotto wrapper of Cellpose to perform segmentation gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom)) writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T) #Create a giotto image to evaluate segmentation gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F) Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'), mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), channel_1 = 0, channel_2 = 0, model_name = 'cyto3', batch_size=12) cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'), shift_horizontal_step = F, shift_vertical_step = F, flip_vertical = T) plot(gimg_for_cellpose) plot(cpoly, add = T, border = 'red') 11.4 Create a Giotto Object using list of gitto large images and polygons You will need to have - list of giotto images - giottoPolygon created from segmentation Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer), images = gimg_list, instructions = instrs) Lunaphore_giotto 11.4.1 Overlap to matrix calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with Lunaphore_giotto = calculateOverlap(Lunaphore_giotto, spatial_info = 'cell', image_names = names(gimg_list)) Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto, type ='intensity', poly_info = "cell", feat_info = "protein", aggr_function = "sum") showGiottoExpression(Lunaphore_giotto) 11.4.2 Manipulate Expression information For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile. Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix') filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',] Lunaphore_giotto <- setExpression(Lunaphore_giotto, feat_type = 'protein', x = createExprObj(filtered_expr_mtx), name = 'raw') showGiottoExpression(Lunaphore_giotto) 11.4.3 Rescale polygons rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto, poly_info = 'cell', name = 'smallcell', fx = 0.7, fy = 0.7, calculate_centroids = T) smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell') plot(gimg_DAPI,ext = zoom) plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom) plot(smallpoly,add = TRUE, border = "red",ext = zoom) 11.4.4 Perform clustering and differential expression The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein') Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein', scale_unit = F, center = F, ncp = 20,feats_to_use = NULL, set_seed = TRUE) screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T) Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check. Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5, set_seed = TRUE) Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5) Lunaphore_giotto = doHclust(Lunaphore_giotto, k = 8, dim_reduction_to_use = 'cells', spat_unit = 'cell', feat_type = 'protein') spatInSituPlotPoints(gobject = Lunaphore_giotto, spat_unit = "cell", polygon_feat_type = "cell", show_polygon = T, feat_type = "protein", feats = NULL, polygon_fill = 'hclust', polygon_fill_as_factor = TRUE, polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T, polygon_color = 'black', background_color = "white") Then we can check the heatmap of protein expression and determine the first round of cluster annotation cluster_column = 'hclust' plotMetaDataHeatmap(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', expression_values = "raw", metadata_cols = c(cluster_column), selected_feats = names(gimg_list), y_text_size = 8, show_values = 'zscores_rescaled') 11.4.5 Give the cluster an annotation based on expression values annotation = c('B_cell','Macrophage','T_cell','stromal','epithelial','DC','Fibroblast','endothelial') names(annotation) = 1:8 Lunaphore_giotto <- annotateGiotto(Lunaphore_giotto, cluster_column = 'hclust', annotation_vector = annotation, name = "cell_types") 11.4.6 spatial network This is to create a cellular neighborhood based on nearest neighbor of physical distance Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto) spatPlot2D(Lunaphore_giotto, show_network = T, network_color = 'blue', point_size = 1.5, cell_color = 'hclust') 11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions This is using cellProximityEnrichment() to statistically identify cell type interactions cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto, cluster_column = 'cell_types', spatial_network_name = 'Delaunay_network', adjust_method = 'fdr', number_of_simulations = 2000) ## barplot cellProximityBarplot(gobject = Lunaphore_giotto, CPscore = cell_proximities, min_orig_ints = 5, min_sim_ints = 5) ## network cellProximityNetwork(gobject = Lunaphore_giotto, CPscore = cell_proximities, remove_self_edges = T, only_show_enrichment_edges = F) "],["working-with-multiple-samples.html", "12 Working with multiple samples 12.1 Objective 12.2 Background 12.3 Create individual giotto objects 12.4 Join Giotto Objects 12.5 Visualizing combined datasets 12.6 Splitting combined dataset 12.7 Analyzing joined objects 12.8 Perform Harmony and default workflows", " 12 Working with multiple samples Jeff Sheridan August 7th 2024 12.1 Objective Giotto enables the grouping of multiple objects into a single object for combined analysis. Datasets can be spatially distributed across the x, y, or z axes, allowing for the creation of 3D datasets using the z-plane or the analysis of grouped datasets, such as multiple replicates or similar samples. While it’s possible to integrate multiple datasets, batch effects from samples can hinder effective integration. In such cases, more sophisticated methods may be needed to successfully integrate and cluster samples as a unified dataset. One example of an advanced integration technique is Harmony, which will be discussed in more detail later in this tutorial. This tutorial will demonstrate the integration of two Visium datasets, examining the results before and after Harmony integration. 12.2 Background 12.2.1 Dataset For this tutorial we will be using two prostate visium datasets produced by 10X Genomics, one an Adenocarcinoma with Invasive Carcinoma and the other a normal prostate sample. 12.2.2 Visium technology Figure 12.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 12.3 Create individual giotto objects 12.3.1 Download the data You need to download the expression matrix and spatial information by running these commands: data_dir <- "data/3_session1" dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), showWarnings = TRUE, recursive = TRUE) # Spatial data adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz")) # Download matrix adenocarcinoma prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz")) dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), showWarnings = FALSE, recursive = TRUE) # Spatial data normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz")) # Download matrix normal prostate download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz", destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz")) 12.3.2 Create giotto instructions We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances. library(Giotto) save_dir <- "results/03_session1" instrs <- createGiottoInstructions(save_dir = save_dir, save_plot = TRUE, show_plot = TRUE, python_path = NULL) 12.3.3 Load visium data into Giotto We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image. print(data_dir) print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate")) ## Healthy prostate N_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) ## Adenocarcinoma C_pros <- createGiottoVisiumObject( visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"), expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instrs ) We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image. Figure 12.2: Structure of Giotto object containing a single dataset. 12.3.4 Healthy prostate tissue coverage Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”. spatPlot2D(gobject = N_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_normal_prostate_tissue")) Figure 12.3: Tissue coverage for the normal prostate sample. 12.3.5 Adenocarcinoma prostate tissue coverage spatPlot2D(gobject = C_pros, cell_color = "in_tissue", show_image = TRUE, point_size = 2.5, cell_color_code = c("black", "red"), point_alpha = 0.5, save_param = list(save_name = "03_ses1_adeno_prostate_tissue")) Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. 12.3.6 Showing the data strucutre for the inidividual objects # Printing the file structure for the individual datasets print(head(pDataDT(N_pros))) print(N_pros) 12.4 Join Giotto Objects To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image. combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros), gobject_names = c("NP", "CP"), join_method = "shift", x_padding = 1000) # Printing the file structure for the individual datasets print(head(pDataDT(combined_pros))) print(combined_pros) From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”. Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.5 Visualizing combined datasets The combined dataset can either visualized in the same space or in two separate plots through the group_by variable. To show images both the show_image variable and the image_name variable containing both image names needs to be used. 12.5.1 Vizualizing in the same plot Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, image_name = c("NP-image", "CP-image"), point_size = 1, point_alpha = 0.5, save_param = list(save_name = "03_ses1_combined_tissue")) Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. 12.5.2 Vizualizing on separate plots If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable. spatPlot2D(gobject = combined_pros, cell_color = "in_tissue", cell_color_code = c("black", "pink"), show_image = TRUE, image_name = c("NP-image", "CP-image"), group_by = "list_ID", point_alpha = 0.5, point_size = 0.5, cow_n_col = 1, save_param = list(save_name = "03_ses1_combined_tissue_group")) Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. 12.6 Splitting combined dataset If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below. # Getting the cell information combined_cells <- pDataDT(combined_pros) np_cells <- combined_cells[list_ID == "NP"] np_split <- subsetGiotto(combined_pros, cell_ids = np_cells$cell_ID, poly_info = np_cells$cell_ID, spat_unit = "cell") spatPlot2D(gobject = np_split, cell_color = "in_tissue", cell_color_code = c("black", "red"), show_image = TRUE, point_alpha = 0.5, point_size = 0.5, save_param = list(save_name = "03_ses1_split_object")) Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. 12.7 Analyzing joined objects 12.7.1 Normalization and adding statistics Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets. # subset on in-tissue spots metadata <- pDataDT(combined_pros) in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID combined_pros <- subsetGiotto(combined_pros, cell_ids = in_tissue_barcodes) ## filter combined_pros <- filterGiotto(gobject = combined_pros, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 500, expression_values = "raw", verbose = TRUE) ## normalize combined_pros <- normalizeGiotto(gobject = combined_pros, scalefactor = 6000) ## add gene & cell statistics combined_pros <- addStatistics(gobject = combined_pros, expression_values = "raw") ## visualize spatPlot2D(gobject = combined_pros, cell_color = "nr_feats", color_as_factor = FALSE, point_size = 1, show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_feat_expression")) After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples. Figure 12.9: Unique feat expression for visium spots for both prostate samples. 12.7.2 Clustering the datasets Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together ## PCA ## combined_pros <- calculateHVF(gobject = combined_pros) combined_pros <- runPCA(gobject = combined_pros, center = TRUE, scale_unit = TRUE) ## cluster and run UMAP ## # sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "pca", dim_reduction_name = "pca", dimensions_to_use = 1:10, k = 15) # Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, resolution = 0.2, n_iterations = 200) # UMAP combined_pros <- runUMAP(combined_pros) 12.7.3 Vizualizing spatial location of clusters We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot. spatDimPlot2D(gobject = combined_pros, cell_color = "leiden_clus", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_leiden_clus")) Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.7.4 Vizualizing tissue contribution to clusters We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples. spatDimPlot2D(gobject = combined_pros, cell_color = "list_ID", show_image = TRUE, image_name = c("NP-image", "CP-image"), save_param = list(save_name = "ses3_1_tissue_contribution")) Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8 Perform Harmony and default workflows We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes. Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations. library(harmony) ## run harmony integration combined_pros <- runGiottoHarmony(combined_pros, vars_use = "list_ID") After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps. Figure 12.12: Data structure of the gobject after running Harmony integration. 12.8.1 Clustering harmonized object We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis. ## sNN network (default) combined_pros <- createNearestNetwork(gobject = combined_pros, dim_reduction_to_use = "harmony", dim_reduction_name = "harmony", name = "NN.harmony", dimensions_to_use = 1:10, k = 15) ## Leiden clustering combined_pros <- doLeidenCluster(gobject = combined_pros, network_name = "NN.harmony", resolution = 0.2, n_iterations = 1000, name = "leiden_harmony") # UMAP dimension reduction combined_pros <- runUMAP(combined_pros, dim_reduction_name = "harmony", dim_reduction_to_use = "harmony", name = "umap_harmony") spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "leiden_harmony", show_image = TRUE, image_name = c("NP-image", "CP-image"), spat_point_size = 1, save_param = list(save_name = "leiden_clustering_harmony")) We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially. Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). 12.8.2 Vizualizing the tissue contribution We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue. spatDimPlot2D(gobject = combined_pros, dim_reduction_to_use = "umap", dim_reduction_name = "umap_harmony", cell_color = "list_ID", save_plot = TRUE, save_param = list(save_name = "leiden_clustering_harmony_contribution")) Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). "],["spatial-multi-modal-analysis.html", "13 Spatial multi-modal analysis 13.1 Overview 13.2 Spatial manipulation 13.3 Examples of the simple transforms with a giottoPolygon 13.4 Affine transforms 13.5 Image transforms 13.6 The practical usage of multi-modality co-registration", " 13 Spatial multi-modal analysis George Chen Junxiang Xu August 7th 2024 13.1 Overview Spatial multimodal datasets are created when there is more than one modality available for a single piece of tissue. One way that these datasets can be assembled is by performing multiple spatial assays on closely adjacent tissue sections or ideally the same section. However, for these datasets, in addition to the usual expression space integration, we must also first spatially align them. 13.2 Spatial manipulation Performing spatial analyses across any two sections of tissue from the same block requires that data to be spatially aligned into a common coordinate space. Minute differences during the sectioning process from the cutting motion to how long an FFPE section was floated can result in even neighboring sections being distorted when compared side-by-side. These differences make it difficult to assemble multislice and/or cross platform multimodal datasets into a cohesive 3D volume. The solution for this is to perform registration across either the dataset images or expression information. Based on the registration results, both the raster images and vector feature and polygon information can be aligned into a continuous whole. Ideally this registration will be a free deformation based on sets of control points or a deformation matrix, however affine transforms already provide a good approximation. In either case, the transform or deformation applied must work in the same way across both raster and vector information. Giotto provides spatial classes and methods for easy manipulation of data with 2D affine transformations. These functionalities are all available from GiottoClass. 13.2.1 Spatial transforms: We support simple transformations and more complex affine transformations which can be used to combine and encode more than one simple transform. spatShift() - translations spin() - rotations (degrees) rescale() - scaling flip() - flip vertical or horizontal across arbitrary lines t() - transpose shear() - shear transform affine() - affine matrix transform 13.2.2 Spatial utilities: Helpful functions for use alongside these spatial transforms are ext() for finding the spatial bounding box of where your data object is, crop() for cutting out a spatial region of the data, and plot() for terra/base plots of the data. ext() - spatial extent or bounding box crop() - cut out a spatial region of the data plot() - plot a spatial object 13.2.3 Spatial classes: Giotto’s spatial subobjects respond to the above functions. The Giotto object itself can also be affine transformed. spatLocsObj - xy centroids spatialNetworkObj - spatial networks between centroids giottoPoints - xy feature point detections giottoPolygon - spatial polygons giottoImage (mostly deprecated) - magick-based images giottoLargeImage/giottoAffineImage - terra-based images affine2d - affine matrix container giotto - giotto analysis object # load in data library(Giotto) g <- GiottoData::loadGiottoMini("vizgen") activeSpatUnit(g) <- "aggregate" gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE) gimg <- getGiottoImage(g) 13.3 Examples of the simple transforms with a giottoPolygon rain <- rainbow(nrow(gpoly)) line_width <- 0.3 # par to setup the grid plotting layout p <- par(no.readonly = TRUE) par(mfrow=c(3,3)) gpoly |> plot(main = "no transform", col = rain, lwd = line_width) gpoly |> spatShift(dx = 1000) |> plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width) gpoly |> spin(45) |> plot(main = "spin(45)", col = rain, lwd = line_width) gpoly |> rescale(fx = 10, fy = 6) |> plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width) gpoly |> flip(direction = "vertical") |> plot(main = "flip()", col = rain, lwd = line_width) gpoly |> t() |> plot(main = "t()", col = rain, lwd = line_width) gpoly |> shear(fx = 0.5) |> plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width) par(p) 13.4 Affine transforms The above transforms are all simple to understand in how they work, but you can imagine that performing them in sequence on your dataset can be computationally expensive. Luckily, the above operations are all affine transformation, and they can be condensed into a single step. Affine transforms where the x and y values undergo a linear transform. These transforms in 2D, can all be represented as a 2x2 matrix or 2x3 if the xy translation values are included. To perform the linear transform, the xy coordinates just need to be matrix multiplied by the 2x2 affine matrix. The resulting values should then be added to the translate values. Due to the nature of matrix multiplication, you can simply multiply the affine matrices with each other and when the xy coordinates are multiplied by the resulting matrix, it performs both linear transforms in the same step. Giotto provides a utility affine2d S4 class that can be created from any affine matrix and responds to the affine transform functions to simplify this accumulation of simple transforms. Once done, the affine2d can be applied to spatial objects in a single step using affine() in the same way that you would use a matrix. # create affine2d aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1))) The affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set. aff@anchor <- ext(gpoly) aff <- initialize(aff) # append several simple transforms aff <- aff |> spatShift(dx = 1000) |> spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used flip(direction = "vertical") |> t() |> shear(fx = 0.5) force(aff) <affine2d> anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax) rotate : -0.785398163397448 (rad) shear : 0.5, 0 (x, y) scale : 10, 10 (x, y) translate : 963.028150700062, 7071.06781186548 (x, y) The show() function displays some information about the stored affine transform, including a set of decomposed simple transformations. You can then plot the affine object and see a projection of the spatial transform where blue is the starting position and red is the end. plot(aff) We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests. gpoly |> affine(aff) |> plot(main = "affine()", col = rain, lwd = line_width) 13.5 Image transforms Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting. spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image. # affine transform of image info matches with polygon info gimg |> affine(aff) |> plot() gpoly |> affine(aff) |> plot(add = TRUE, border = "cyan", lwd = 0.3) # affine of the giotto object g |> affine(aff) |> spatInSituPlotPoints( show_image = TRUE, feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")), feats_color_code = rainbow(4), polygon_color = "cyan", polygon_line_size = 0.1, point_size = 0.1, use_overlap = FALSE ) Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used. 13.6 The practical usage of multi-modality co-registration 13.6.1 Example dataset: Xenium Breast Cancer pre-release pack 10X Genomics Released a comprehensive dataset on 2022. To capture spatial structure by complementing different spatial resolutions and modalities across different assays, they provided a dataset with Xenium in situ transcriptomics data, together with Visium on closely adjacent sections. Additional IF staining was also performed on the Xenium slides. For more information, please refer to the pre-release dataset page as well as the publication. Visium – H&E Histology – 55um spot level expression with transcriptome coverage Xenium – H&E Histology – IF image staining DAPI, HER2 and CD20 – in situ transcripts – cooresponding centroid locations The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance. library(Giotto) instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'), save_plot = TRUE, show_plot = TRUE) options(timeout = 999999) download_dir <-file.path(getwd(),'/data/03_session2/') destfile <- file.path(download_dir,'Multimodal_registration.zip') if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) } download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile) unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir) Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/') Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/') 13.6.2 Target Coordinate system Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system. xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz")) xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() xen_cell_pl 13.6.3 Visium to register Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir, gene_column_index = 2, png_name = 'tissue_hires_image.png', instructions = NULL) # In the meantime, calculate statistics for easier plot showing G_visium <- normalizeGiotto(G_visium) G_visium <- addStatistics(G_visium) V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T) V_origin The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform. # create affine2d aff <- affine(diag(c(1,1))) aff <- aff |> spin(90) |> flip(direction = "horizontal") force(aff) # Apply the transform V_tansformed <- affine(G_visium,aff) spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T) spatplot_to_register Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app. landmarks <- interactiveLandmarkSelection(spatplot_to_register, xen_cell_pl) Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result. landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds')) affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]]) V_final <- affine(G_visium,affine_mtx %*% aff@affine) spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F) spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.3.1 Create Pseudo Visium dataset for comparison Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance. V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN') spat_network <- getSpatialNetwork(V_final,output = 'networkDT') spatPlot2D(V_final, show_network = T, network_color = 'blue', point_size = 1) center_to_center <- min(spat_network$distance) radius <- center_to_center*55/200 Now we get the Pseudo Visium polygons Visium_centroid <- getSpatialLocations(V_final,output = 'data.table') stamp_dt <- circleVertices(radius = radius, npoints = 100) pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid) pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T) plot(pseudo_visium_poly) Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz')) gpoints <- createGiottoPoints(xen_transcripts) Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints), gpolygons = list('visium' = pseudo_visium_poly)) Get gene expression information by overlapping polygon to points Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'visium') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "visium", feat_info = "rna", aggr_function = "sum") Manipulate the expression for plotting Xen_obj <- filterGiotto(Xen_obj, feat_type = 'rna', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, point_size = 3.5, expression_values = 'plot', show_image = F, feats = 'ERBB2') Subset the registered Visium and plot same gene #get the extent of giotto points, xmin, xmax, ymin, ymax subset_extent <- ext(gpoints@spatVector) sub_visium <- subsetGiottoLocs(V_final, x_min = subset_extent[1], x_max = subset_extent[2], y_min = subset_extent[3], y_max = subset_extent[4]) spatFeatPlot2D(sub_visium, point_size = 2, expression_values = 'scaled', show_image = F, feats = 'ERBB2') 13.6.4 Register post-Xenium H&E and IF image For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics. HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"), imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"), micron = 0.2125) plot(HE_xen) The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use. HE_xen <- HE_xen |> flip(direction = "vertical") gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen))) plot(gimg_rast) #terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info) Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots gg <- ggplot2::ggplot() pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast) pl + ggplot2::geom_smooth() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic() 13.6.4.1 Add registered image information and compare RNA vs protein expression With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20') HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2') Xen_obj <- addGiottoLargeImage(gobject = Xen_obj, largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg)) Get the cell polygons, as Xenium and IF are both subcellular resolution cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz')) colnames(cellpoly_dt) <- c('poly_ID','x','y') cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt) Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly)) Compute the gene expression matrix by overlay the cell polygons and giotto points. Xen_obj <- calculateOverlap(Xen_obj, feat_info = 'rna', spatial_info = 'cell') Xen_obj <- overlapToMatrix(x = Xen_obj, type = "point", poly_info = "cell", feat_info = "rna", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'rna', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'rna', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'rna', expression_values = 'plot', spat_unit = 'cell', feats = 'ERBB2', point_size = 0.05) Now we overlay the HER2 expression from the raster image with the cell polygons. Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'cell', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "cell", feat_info = "protein", aggr_function = "sum") tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'cell', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'cell', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'cell', feats = 'HER2', point_size = 0.05) We can also overlay the protein expression to Visium spots Xen_obj <- calculateOverlap(Xen_obj, spatial_info = 'visium', image_names = c('HER2','CD20')) Xen_obj <- overlapToMatrix(x = Xen_obj, type = "intensity", poly_info = "visium", feat_info = "protein", aggr_function = "sum") Xen_obj <- filterGiotto(Xen_obj, feat_type = 'protein', spat_unit = 'visium', expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) tmp_exprs <- getExpression(Xen_obj, feat_type = 'protein', spat_unit = 'visium', output = 'matrix') Xen_obj <- setExpression(Xen_obj, x = createExprObj(log(tmp_exprs+1)), feat_type = 'protein', spat_unit = 'visium', name = 'plot') spatFeatPlot2D(Xen_obj, feat_type = 'protein', expression_values = 'plot', spat_unit = 'visium', feats = 'HER2', point_size = 2) 13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially. Here, we provide an example of two compressed images to show the automatic alignment pipeline. HE <- createGiottoLargeImage(paste0(Xenium_dir,'mini_HE.png'),negative_y = F) plot(HE) IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T) terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin") Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform. estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'), flip_horizontal = T, use_single_channel = T, single_channel_number = 3), y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'), invert = T), plot_match = T, max_ratio = 0.5,estimate_fun = 'Projective') Use the estimation, we can quickly visualize the transformation mtx <- as.matrix(estimation$params) transformed <- affine(IF, mtx) To_see_overlay <- transformed@funs$realize_magick(size = 2e6) plot(HE) plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5) SIFT_registration_overlay 13.6.6 Final Notes Image registration is becoming crucial for spatial multi modal analysis. The methods included here are not the only ways to register images, and either of them may have drawbacks for a good alignment. There are multiple tools coming out for the field with different strategies, including easier landmark detection, deformable transformation as well as matching spatial patterns, etc. Some of them provides transformed images or coordinates that can be directly loaded to Giotto as a multimodal object using a standard pipeline. "],["multi-omics-integration.html", "14 Multi-omics integration 14.1 The CytAssist technology 14.2 Introduction to the spatial dataset 14.3 Download dataset 14.4 Create the Giotto object 14.5 Subset on spots that were covered by tissue 14.6 RNA processing 14.7 Protein processing 14.8 Multi-omics integration 14.9 Session info", " 14 Multi-omics integration Joselyn Cristina Chávez Fuentes August 7th 2024 14.1 The CytAssist technology The Visium CytAssist Spatial Gene and Protein Expression assay is designed to introduce simultaneous Gene Expression and Protein Expression analysis to FFPE samples processed with Visium CytAssist. The assay uses NGS to measure the abundance of oligo-tagged antibodies with spatial resolution, in addition to the whole transcriptome and a morphological image. Figure 14.1: CystAssits multi-omics diagram. Source: 10X genomics. The 10X human immune cell profiling panel features 35 antibodies from Abcam and Biolegend, and includes cell surface and intracellular targets. The rna probes hybridize to ~18,000 genes, or RNA targets, within the tissue section to achieve whole transcriptome gene expression profiling. The remaining steps, starting with probe extension, follow the standard Visium workflow outside of the instrument. Figure 14.2: CytAssist workflow. Source: 10X genomics. 14.2 Introduction to the spatial dataset The Human Tonsil (FFPE) dataset was obtained from 10X Genomics. The tissue was sectioned as described in Visium CytAssist Spatial Gene and Protein Expression for FFPE – Tissue Preparation Guide (CG000660). 5 µm tissue sections were placed on Superfrost glass slides, deparaffinized, H&E stained (CG000658) and coverslipped. Sections were imaged, decoverslipped, followed by decrosslinking per the Staining Demonstrated Protocol (CG000658). More information about this dataset can be found here. 14.3 Download dataset You need to download the expression matrix and spatial information by running these commands: dir.create("data/03_session3") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/2.1.0/CytAssist_FFPE_Protein_Expression_Human_Tonsil/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", destfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/03_session3/”. untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_raw_feature_bc_matrix.tar.gz", exdir = "data/03_session3") untar(tarfile = "data/03_session3/CytAssist_FFPE_Protein_Expression_Human_Tonsil_spatial.tar.gz", exdir = "data/03_session3") 14.4 Create the Giotto object The minimum requirements are: matrix with expression information (or the path to) x,y(,z) coordinates for cells or spots (or the path to) createGiottoVisiumObject() will automatically detect both RNA and Protein modalities in the expression matrix and will create a multi-omics Giotto object. library(Giotto) ## Set instructions results_folder <- "results/03_session3/" python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) # Provide the path to the data folder data_path <- "data/03_session3/" # Create object directly from the data folder visium_tonsil <- createGiottoVisiumObject( visium_dir = data_path, expr_data = "raw", png_name = "tissue_lowres_image.png", gene_column_index = 2, instructions = instructions ) Print the information of the object, note that both rna and protein are listed in the expression slot. visium_tonsil 14.5 Subset on spots that were covered by tissue spatPlot2D( gobject = visium_tonsil, cell_color = "in_tissue", point_size = 2, cell_color_code = c("0" = "lightgrey", "1" = "blue"), show_image = TRUE, image_name = "image" ) Figure 14.3: Spatial plot of the CytAssist human tonsil sample, color indicates wheter the spot is in tissue (1) or not (0). Use the metadata table to identify the spots corresponding to the tissue area, given by the “in_tissue” column. Then use the spot IDs to subset the giotto object. metadata <- getCellMetadata(gobject = visium_tonsil, output = "data.table") in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID visium_tonsil <- subsetGiotto(visium_tonsil, cell_ids = in_tissue_barcodes) 14.6 RNA processing Run the Filtering, normalization, and statistics steps using only the RNA feature. visium_tonsil <- filterGiotto( gobject = visium_tonsil, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil) Dimension reduction Identify the highly variable features using the RNA features, then calculate the principal components based on the HVFs. visium_tonsil <- calculateHVF(gobject = visium_tonsil) visium_tonsil <- runPCA(gobject = visium_tonsil) Clustering Calculate the UMAP, tSNE, and shared nearest neighbor network using the first 10 principal components for the RNA modality. visium_tonsil <- runUMAP(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, dimensions_to_use = 1:10, k = 30) Calculate the RNA-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, resolution = 1, n_iterations = 1000) Visualization Plot the RNA-based UMAP with the corresponding RNA-based Leiden cluster per spot. plotUMAP(gobject = visium_tonsil, cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.4: RNA UMAP, color indicates the RNA-based Leiden clusters. Plot the spatial distribution of the RNA-based Leiden cluster per spot. spatPlot2D(gobject = visium_tonsil, show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.5: Spatial distribution of RNA-based Leiden clusters. 14.7 Protein processing Run the Filtering, normalization, and statistics steps for the protein modality. visium_tonsil <- filterGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1, expression_values = "raw", verbose = TRUE) visium_tonsil <- normalizeGiotto(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", scalefactor = 6000, verbose = TRUE) visium_tonsil <- addStatistics(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Dimension reduction Calculate the principal components using all the proteins available in the dataset. visium_tonsil <- runPCA(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein") Clustering Calculate the UMAP, tSNE, and shared nearest neighbors network using the first 10 principal components for the Protein modality. visium_tonsil <- runUMAP(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- runtSNE(visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10) visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", dimensions_to_use = 1:10, k = 30) Calculate the Protein-based Leiden clusters. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", resolution = 1, n_iterations = 1000) Visualization Plot the Protein UMAP and color the spots using the Protein-based Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", cell_color = "leiden_clus", show_NN_network = TRUE, point_size = 2) Figure 14.6: Protein UMAP, color indicates the Protein-based Leiden clusters. Plot the spatial distribution of the Protein-based Leiden clusters. spatPlot2D(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", show_image = TRUE, cell_color = "leiden_clus", point_size = 3) Figure 14.7: Spatial distribution of Protein-based Leiden clusters. 14.8 Multi-omics integration Calculate kNN Calculate the k nearest neighbors network for each modality (RNA and Protein), using the first 10 principal components of each feature type. ## RNA modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, type = "kNN", dimensions_to_use = 1:10, k = 20) ## Protein modality visium_tonsil <- createNearestNetwork(gobject = visium_tonsil, spat_unit = "cell", feat_type = "protein", type = "kNN", dimensions_to_use = 1:10, k = 20) Run WNN Run the Weighted Nearest Neighbor analysis to weight the contribution of each feature type per spot. The results will be saved in the multiomics slot of the giotto object. visium_tonsil <- runWNN(visium_tonsil, spat_unit = "cell", modality_1 = "rna", modality_2 = "protein", pca_name_modality_1 = "pca", pca_name_modality_2 = "protein.pca", k = 20, integrated_feat_type = NULL, matrix_result_name = NULL, w_name_modality_1 = NULL, w_name_modality_2 = NULL, verbose = TRUE) Run Integrated umap Calculate the UMAP using the weights of each feature per spot. visium_tonsil <- runIntegratedUMAP(visium_tonsil, modality1 = "rna", modality2 = "protein", spread = 7, min_dist = 1, force = FALSE) Calculate integrated clusters Calculate the multiomics-based Leiden clusters using the weights of each feature per spot. visium_tonsil <- doLeidenCluster(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", nn_network_to_use = "kNN", network_name = "integrated_kNN", name = "integrated_leiden_clus", resolution = 1) Visualize the integrated UMAP Plot the integrated UMAP and color the spots using the integrated Leiden clusters. plotUMAP(gobject = visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", dim_reduction_name = "integrated.umap", point_size = 2, title = "Integrated UMAP using Integrated Leiden clusters") Figure 14.8: Integrated UMAP. Color represents the integrated Leiden clusters. Visualize spatial plot with integrated clusters Plot the spatial distribution of the integrated Leiden clusters. spatPlot2D(visium_tonsil, spat_unit = "cell", feat_type = "rna", cell_color = "integrated_leiden_clus", point_size = 3, show_image = TRUE, title = "Integrated Leiden clustering") Figure 14.9: Spatial distribution of the integrated Leiden clusters. 14.9 Session info sessionInfo() R version 4.4.1 (2024-06-14) Platform: aarch64-apple-darwin20 Running under: macOS Sonoma 14.5 Matrix products: default BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0 locale: [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 time zone: America/New_York tzcode source: internal attached base packages: [1] stats graphics grDevices utils datasets methods base other attached packages: [1] Giotto_4.1.0 GiottoClass_0.3.3 loaded via a namespace (and not attached): [1] colorRamp2_0.1.0 deldir_2.0-4 [3] rlang_1.1.4 magrittr_2.0.3 [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10 [7] matrixStats_1.3.0 compiler_4.4.1 [9] png_0.1-8 systemfonts_1.1.0 [11] vctrs_0.6.5 reshape2_1.4.4 [13] stringr_1.5.1 pkgconfig_2.0.3 [15] SpatialExperiment_1.14.0 crayon_1.5.3 [17] fastmap_1.2.0 backports_1.5.0 [19] magick_2.8.4 XVector_0.44.0 [21] labeling_0.4.3 utf8_1.2.4 [23] rmarkdown_2.27 UCSC.utils_1.0.0 [25] ragg_1.3.2 purrr_1.0.2 [27] xfun_0.46 beachmat_2.20.0 [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1 [31] jsonlite_1.8.8 DelayedArray_0.30.1 [33] BiocParallel_1.38.0 terra_1.7-78 [35] irlba_2.3.5.1 parallel_4.4.1 [37] R6_2.5.1 stringi_1.8.4 [39] RColorBrewer_1.1-3 reticulate_1.38.0 [41] parallelly_1.37.1 GenomicRanges_1.56.1 [43] scattermore_1.2 Rcpp_1.0.13 [45] bookdown_0.40 SummarizedExperiment_1.34.0 [47] knitr_1.48 future.apply_1.11.2 [49] R.utils_2.12.3 IRanges_2.38.1 [51] Matrix_1.7-0 igraph_2.0.3 [53] tidyselect_1.2.1 rstudioapi_0.16.0 [55] abind_1.4-5 yaml_2.3.9 [57] codetools_0.2-20 listenv_0.9.1 [59] lattice_0.22-6 tibble_3.2.1 [61] plyr_1.8.9 Biobase_2.64.0 [63] withr_3.0.0 Rtsne_0.17 [65] evaluate_0.24.0 future_1.33.2 [67] pillar_1.9.0 MatrixGenerics_1.16.0 [69] checkmate_2.3.1 stats4_4.4.1 [71] plotly_4.10.4 generics_0.1.3 [73] dbscan_1.2-0 sp_2.1-4 [75] S4Vectors_0.42.1 ggplot2_3.5.1 [77] munsell_0.5.1 scales_1.3.0 [79] globals_0.16.3 gtools_3.9.5 [81] glue_1.7.0 lazyeval_0.2.2 [83] tools_4.4.1 GiottoVisuals_0.2.4 [85] data.table_1.15.4 ScaledMatrix_1.12.0 [87] cowplot_1.1.3 grid_4.4.1 [89] tidyr_1.3.1 colorspace_2.1-0 [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12 [93] BiocSingular_1.20.0 rsvd_1.0.5 [95] cli_3.6.3 textshaping_0.4.0 [97] fansi_1.0.6 S4Arrays_1.4.1 [99] viridisLite_0.4.2 dplyr_1.1.4 [101] uwot_0.2.2 gtable_0.3.5 [103] R.methodsS3_1.8.2 digest_0.6.36 [105] BiocGenerics_0.50.0 SparseArray_1.4.8 [107] ggrepel_0.9.5 farver_2.1.2 [109] rjson_0.2.21 htmlwidgets_1.6.4 [111] htmltools_0.5.8.1 R.oo_1.26.0 [113] lifecycle_1.0.4 httr_1.4.7 "],["interoperability-with-other-frameworks.html", "15 Interoperability with other frameworks 15.1 Load Giotto object 15.2 Seurat 15.3 SpatialExperiment 15.4 AnnData 15.5 Create mini Vizgen object", " 15 Interoperability with other frameworks Iqra August 7th 2024 Giotto facilitates seamless interoperability with various tools, including Seurat, AnnData, and SpatialExperiment. Below is a comprehensive tutorial on how Giotto interoperates with these other tools. 15.1 Load Giotto object To begin demonstrating the interoperability of a Giotto object with other frameworks, we first load the required libraries and a Giotto mini object. We then proceed with the conversion process: library(Giotto) library(GiottoData) Load a Giotto mini Visium object, which will be used for demonstrating interoperability. gobject <- GiottoData::loadGiottoMini("visium") 15.2 Seurat Giotto Suite provides interoperability between Seurat and Giotto, supporting both older and newer versions of Seurat objects. The four tailored functions are giottoToSeuratV4(), seuratToGiottoV4() for older versions, and giottoToSeuratV5(), seuratToGiottoV5() for Seurat v5, which includes subcellular and image information. These functions map Giotto’s metadata, dimension reductions, spatial locations, and images to the corresponding slots in Seurat. 15.2.1 Conversion of Giotto Object to Seurat Object To convert Giotto object to Seurat V5 object, we first load required libraries and use the function giottoToSeuratV5() function library(Seurat) library(SeuratData) library(ggplot2) library(patchwork) library(dplyr) Now we convert the Giotto object to a Seurat V5 object and create violin and spatial feature plots to visualize the RNA count data. gToS <- giottoToSeuratV5(gobject = gobject, spat_unit = "cell") plot1 <- VlnPlot(gToS, features = "nCount_rna", pt.size = 0.1) + NoLegend() plot2 <- SpatialFeaturePlot(gToS, features = "nCount_rna", pt.size.factor = 2) + theme(legend.position = "right") wrap_plots(plot1, plot2) 15.2.1.1 Apply SCTransform We apply SCTransform to perform data transformation on the RNA assay: SCTransform() function. gToS <- SCTransform(gToS, assay = "rna", verbose = FALSE) 15.2.1.2 Dimension Reduction We perform Principal Component Analysis (PCA), find neighbors, and run UMAP for dimensionality reduction and clustering on the transformed Seurat object: gToS <- RunPCA(gToS, assay = "SCT") gToS <- FindNeighbors(gToS, reduction = "pca", dims = 1:30) gToS <- RunUMAP(gToS, reduction = "pca", dims = 1:30) 15.2.2 Conversion of Seurat object Back to Giotto Object To Convert the Seurat Object back to Giotto object, we use the funcion seuratToGiottoV5(), specifying the spatial assay, dimensionality reduction techniques, and spatial and nearest neighbor networks. giottoFromSeurat <- seuratToGiottoV5(sobject = gToS, spatial_assay = "rna", dim_reduction = c("pca", "umap"), sp_network = "Delaunay_network", nn_network = c("sNN.pca", "custom_NN" )) 15.2.2.1 Clustering and Plotting UMAP Here we perform K-means clustering on the UMAP results obtained from the Seurat object: ## k-means clustering giottoFromSeurat <- doKmeans(gobject = giottoFromSeurat, dim_reduction_to_use = "pca") #Plot UMAP post-clustering to visualize kmeans graph2 <- Giotto::plotUMAP( gobject = giottoFromSeurat, cell_color = "kmeans", show_NN_network = TRUE, point_size = 2.5 ) 15.2.2.2 Spatial CoExpression We can also use the binSpect function to analyze spatial co-expression using the spatial network Delaunay network from the Seurat object and then visualize the spatial co-expression using the heatmSpatialCorFeat() function: ranktest <- binSpect(giottoFromSeurat, bin_method = "rank", calc_hub = TRUE, hub_min_int = 5, spatial_network_name = "Delaunay_network") ext_spatial_genes <- ranktest[1:300,]$feats spat_cor_netw_DT <- detectSpatialCorFeats( giottoFromSeurat, method = "network", spatial_network_name = "Delaunay_network", subset_feats = ext_spatial_genes) top10_genes <- showSpatialCorFeats(spat_cor_netw_DT, feats = "Dsp", show_top_feats = 10) spat_cor_netw_DT <- clusterSpatialCorFeats(spat_cor_netw_DT, name = "spat_netw_clus", k = 7) heatmSpatialCorFeats( giottoFromSeurat, spatCorObject = spat_cor_netw_DT, use_clus_name = "spat_netw_clus", heatmap_legend_param = list(title = NULL), save_plot = TRUE, show_plot = TRUE, return_plot = FALSE, save_param = list(base_height = 6, base_width = 8, units = 'cm')) 15.3 SpatialExperiment For the Bioconductor group of packages, the SpatialExperiment data container handles data from spatial-omics experiments, including spatial coordinates, images, and metadata. Giotto Suite provides giottoToSpatialExperiment() and spatialExperimentToGiotto(), mapping Giotto’s slots to the corresponding SpatialExperiment slots. Since SpatialExperiment can only store one spatial unit at a time, giottoToSpatialExperiment() returns a list of SpatialExperiment objects, each representing a distinct spatial unit. To start the conversion of a Giotto mini Visium object to a SpatialExperiment object, we first load the required libraries. library(SpatialExperiment) library(ggspavis) library(pheatmap) library(scater) library(scran) library(nnSVG) 15.3.1 Convert Giotto Object to SpatialExperiment Object To convert the Giotto object to a SpatialExperiment object, we use the giottoToSpatialExperiment() function. gspe <- giottoToSpatialExperiment(gobject) The conversion function returns a separate SpatialExperiment object for each spatial unit. We select the first object for downstream use: spe <- gspe[[1]] 15.3.1.1 Identify top spatially variable genes with nnSVG We employ the nnSVG package to identify the top spatially variable genes in our SpatialExperiment object. Covariates can be added to our model; in this example, we use Leiden clustering labels as a covariate: # One of the assays should be "logcounts" # We rename the normalized assay to "logcounts" assayNames(spe)[[2]] <- "logcounts" # Create model matrix for leiden clustering labels X <- model.matrix(~ colData(spe)$leiden_clus) dim(X) Run nnSVG This step will take several minutes to run spe <- nnSVG(spe, X = X) # Show top 10 features rowData(spe)[order(rowData(spe)$rank)[1:10], ]$feat_ID 15.3.2 Conversion of SpatialExperiment object back to Giotto We then convert the processed SpatialExperiment object back into a Giotto object for further downstream analysis using the Giotto suite. This is done using the spatialExperimentToGiotto function, where we explicitly specify the spatial network from the SpatialExperiment object. giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") giottoFromSPE <- spatialExperimentToGiotto(spe = spe, python_path = NULL, sp_network = "Delaunay_network") print(giottoFromSPE) 15.3.2.1 Plotting top genes from nnSVG in Giotto Now, we visualize the genes previously identified in the SpatialExperiment object using the nnSVG package within the Giotto toolkit, leveraging the converted Giotto object. ext_spatial_genes <- getFeatureMetadata(giottoFromSPE, output = "data.table") ext_spatial_genes <- ext_spatial_genes[order(ext_spatial_genes$rank)[1:10], ]$feat_ID spatFeatPlot2D(giottoFromSPE, expression_values = "scaled_rna_cell", feats = ext_spatial_genes[1:4], point_size = 2) 15.4 AnnData The anndataToGiotto() and giottoToAnnData() functions map the slots of the Giotto object to the corresponding locations in a Squidpy-flavored AnnData object. Specifically, Giotto’s expression slot maps to adata.X, spatial_locs to adata.obsm, cell_metadata to adata.obs, feat_metadata to adata.var, dimension_reduction to adata.obsm, and nn_network and spat_network to adata.obsp. Images are currently not mapped between both classes. Notably, Giotto stores expression matrices within separate spatial units and feature types, while AnnData does not support this hierarchical data storage. Consequently, multiple AnnData objects are created from a Giotto object when there are multiple spatial unit and feature type pairs. 15.4.1 Load Required Libraries To start, we need to load the necessary libraries, including reticulate for interfacing with Python. library(reticulate) 15.4.2 Specify Path for Results First, we specify the directory where the results will be saved. Additionally, we retrieve and update Giotto instructions. # Specify path to which results may be saved results_directory <- "results/03_session4/giotto_anndata_conversion/" instrs <- showGiottoInstructions(gobject) mini_gobject <- replaceGiottoInstructions(gobject = gobject, instructions = instrs) 15.4.2.1 Create Default kNN Network We will create a k-nearest neighbor (kNN) network using mostly default parameters. gobject <- createNearestNetwork(gobject = gobject, spat_unit = "cell", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 15, name = "kNN.umap") 15.4.3 Giotto To AnnData To convert the giotto object to AnnData, we use the Giotto’s function giottoToAnnData() gToAnnData <- giottoToAnnData(gobject = gobject, save_directory = results_directory) Next, we import scanpy and perform a series of preprocessing steps on the AnnData object. scanpy <- import("scanpy") adata <- scanpy$read_h5ad("results/03_session4/giotto_anndata_conversion/cell_rna_converted_gobject.h5ad") # Normalize total counts per cell scanpy$pp$normalize_total(adata, target_sum=1e4) # Log-transform the data scanpy$pp$log1p(adata) # Perform PCA scanpy$pp$pca(adata, n_comps=40L) # Compute the neighborhood graph scanpy$pp$neighbors(adata, n_neighbors=10L, n_pcs=40L) # Run UMAP scanpy$tl$umap(adata) # Save the processed AnnData object adata$write("results/03_session4/cell_rna_converted_gobject2.h5ad") processed_file_path <- "results/03_session4/cell_rna_converted_gobject2.h5ad" 15.4.4 Convert AnnData to Giotto Finally, we convert the processed AnnData object back into a Giotto object for further analysis using Giotto. giottoFromAnndata <- anndataToGiotto(anndata_path = processed_file_path) 15.4.4.1 UMAP Visualization Now we plot the UMAP using the GiottoVisuals::plotUMAP() function that was calculated using Scanpy on the AnnData object. Giotto::plotUMAP( gobject = giottoFromAnndata, dim_reduction_name = "umap.ad", cell_color = "leiden_clus", point_size = 3 ) 15.5 Create mini Vizgen object mini_gobject <- loadGiottoMini(dataset = "vizgen", python_path = NULL) mini_gobject <- replaceGiottoInstructions(gobject = mini_gobject, instructions = instrs) mini_gobject <- createNearestNetwork(gobject = mini_gobject, spat_unit = "aggregate", feat_type = "rna", type = "kNN", dim_reduction_to_use = "umap", dim_reduction_name = "umap", k = 6, name = "new_network") Since we have multiple spat_unit and feat_type pairs, this function will create multiple .h5ad files, with their names returned. Non-default nearest or spatial network names will have their key_added terms recorded and saved in corresponding .txt files; refer to the documentation for details. anndata_conversions <- giottoToAnnData(gobject = mini_gobject, save_directory = results_directory, python_path = NULL) "],["interoperability-with-isolated-tools.html", "16 Interoperability with isolated tools 16.1 Spatial niche trajectory analysis", " 16 Interoperability with isolated tools Wen Wang August 7th 2024 16.1 Spatial niche trajectory analysis 16.1.1 Dataset download The MERFISH mouse motor cortex data to run this tutorial can be found here You need to download the processed expression, metadata, and cell segmentation information by running these commands: Notes: there are 61 slices here, we run on two of them to save the time. data_path <- "data" dir.create(data_path) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/counts.h5ad", destfile = file.path(data_path,"counts.h5ad")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/cell_labels.csv", destfile = file.path(data_path,"cell_labels.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample1.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample1.csv")) download.file(url = "https://download.brainimagelibrary.org/cf/1c/cf1c1a431ef8d021/processed_data/segmented_cells_mouse2sample6.csv", destfile = file.path(data_path,"segmented_cells_mouse2sample6.csv")) 16.1.2 Create the Giotto object library(Giotto) library(reticulate) ## Set instructions results_folder <- "results" dir.create(results_folder) python_path <- NULL instructions <- createGiottoInstructions( save_dir = results_folder, save_plot = TRUE, show_plot = FALSE, return_plot = FALSE, python_path = python_path ) ## load data ### meta data meta_df <- read.csv(file.path(data_path, "cell_labels.csv"), colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers colnames(meta_df)[[1]] <- "cell_ID" slice1_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice229",] # subset selected slice meta info slice2_meta_df <- meta_df[meta_df$slice_id == "mouse2_slice300",] # subset selected slice meta info ### counts matrix scanpy_installed <- GiottoClass:::checkPythonPackage("scanpy", env_to_use = "giotto_env") ad2g_path <- system.file("python", "ad2g.py", package = "Giotto") reticulate::source_python(ad2g_path) adata <- read_anndata_from_path(file.path(data_path, "counts.h5ad")) X <- extract_expression(adata) cID <- extract_cell_IDs(adata) fID <- extract_feat_IDs(adata) rownames(X) <- fID colnames(X) <- cID slice1_filter <- cID %in% slice1_meta_df$cell_ID slice2_filter <- cID %in% slice2_meta_df$cell_ID slice1_exp <- X[,slice1_filter] slice2_exp <- X[,slice2_filter] ### cell segmentation to cell location segments_1_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample1.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_2_df <- read.csv(file.path(data_path, "segmented_cells_mouse2sample6.csv"), row.names=1, colClasses = "character") # as the cell IDs are 30 digit numbers, set the type as character to avoid the limitation of R in handling larger integers segments_df <- rbind(segments_1_df, segments_2_df) loc.use <- segments_df[c(slice1_meta_df$cell_ID, slice2_meta_df$cell_ID),] loc.x <- grep('boundaryX_',colnames(loc.use),value = T) loc.y <- grep('boundaryY_',colnames(loc.use),value = T) centr.x <- apply(loc.use[,loc.x],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) centr.y <- apply(loc.use[,loc.y],1,function(x){ temp <- lapply(x,function(y){ as.numeric(unlist(strsplit(y,', '))) }) return (median(unname(unlist(temp)))) }) spatial_locs_df <- data.frame(sdimx = centr.x,sdimy = centr.y) slice1_spatial_locs_df <- spatial_locs_df[slice1_meta_df$cell_ID,] slice2_spatial_locs_df <- spatial_locs_df[slice2_meta_df$cell_ID,] ## giotto obj giotto_obj_1 <- createGiottoObject(expression = slice1_exp, spatial_locs = slice1_spatial_locs_df, cell_metadata = slice1_meta_df) giotto_obj_2 <- createGiottoObject(expression = slice2_exp, spatial_locs = slice2_spatial_locs_df, cell_metadata = slice2_meta_df) giotto_obj = joinGiottoObjects(gobject_list = list(giotto_obj_1, giotto_obj_2), gobject_names = c('mouse2_slice229', 'mouse2_slice300'), # name for each samples join_method = 'z_stack') # We skip the processing process here to save time and use the given cell type # annotation directly ONTraC_input <- getONTraCv1Input(gobject = giotto_obj, cell_type = "subclass", output_path = data_path, spat_unit = "cell", feat_type = "rna", verbose = TRUE) head(ONTraC_input) # Cell_ID Sample x y Cell_Type # <char> <char> <dbl> <dbl> <char> # mouse2_slice229-100101435705986292663283283043431511315 mouse2_slice229 -4828.728 -2203.4502 L6 CT # mouse2_slice229-100104370212612969023746137269354247741 mouse2_slice229 -5405.400 -995.6467 OPC # mouse2_slice229-100128078183217482733448056590230529739 mouse2_slice229 -5731.403 -1071.1735 L2/3 IT # mouse2_slice229-100209662400867003194056898065587980841 mouse2_slice229 -5468.113 -1286.2465 Oligo # mouse2_slice229-100218038012295593766653119076639444055 mouse2_slice229 -6399.986 -959.7440 L2/3 IT # mouse2_slice229-100252992997994275968450436343196667192 mouse2_slice229 -6637.847 -1659.6237 Astro Spatial cell type distributions spatPlot2D(giotto_obj, group_by = "slice_id", cell_color = "subclass", point_size = 1, point_border_stroke = NA, legend_text = 6) 16.1.2.1 Installation ONTraC You could run ONTraC on your own laptop or on an HPC with an NVIDIA GPU node. It will run for less than 10 minutes on this example dataset. For larger datasets, running on an NVIDIA GPU is recommended, otherwise it will take a long time. source ~/.bash_profile conda create -y -n ONTraC python=3.11 conda activate ONTraC pip install git+https://github.com/gyuanlab/ONTraC.git@V2 # Retrieving notices: ...working... done # Channels: # - conda-forge # - bioconda # - r # - pytorch # - defaults # Platform: osx-arm64 # Collecting package metadata (repodata.json): ...working... done # Solving environment: ...working... done # # ## Package Plan ## # # environment location: /Users/wwang/miniconda3/envs/ONTraC # # added / updated specs: # - python=3.11 # # # The following packages will be downloaded: # # package | build # ---------------------------|----------------- # bzip2-1.0.8 | h99b78c6_7 120 KB conda-forge # openssl-3.3.1 | hfb2fe0b_2 2.8 MB conda-forge # setuptools-71.0.4 | pyhd8ed1ab_0 1.4 MB conda-forge # ------------------------------------------------------------ # Total: 4.3 MB # # The following NEW packages will be INSTALLED: # # bzip2 conda-forge/osx-arm64::bzip2-1.0.8-h99b78c6_7 # ca-certificates conda-forge/osx-arm64::ca-certificates-2024.7.4-hf0a4a13_0 # libexpat conda-forge/osx-arm64::libexpat-2.6.2-hebf3989_0 # libffi conda-forge/osx-arm64::libffi-3.4.2-h3422bc3_5 # libsqlite conda-forge/osx-arm64::libsqlite-3.46.0-hfb93653_0 # libzlib conda-forge/osx-arm64::libzlib-1.3.1-hfb2fe0b_1 # ncurses conda-forge/osx-arm64::ncurses-6.5-hb89a1cb_0 # openssl conda-forge/osx-arm64::openssl-3.3.1-hfb2fe0b_2 # pip conda-forge/noarch::pip-24.0-pyhd8ed1ab_0 # python conda-forge/osx-arm64::python-3.11.9-h932a869_0_cpython # readline conda-forge/osx-arm64::readline-8.2-h92ec313_1 # setuptools conda-forge/noarch::setuptools-71.0.4-pyhd8ed1ab_0 # tk conda-forge/osx-arm64::tk-8.6.13-h5083fa2_1 # tzdata conda-forge/noarch::tzdata-2024a-h0c530f3_0 # wheel conda-forge/noarch::wheel-0.43.0-pyhd8ed1ab_1 # xz conda-forge/osx-arm64::xz-5.2.6-h57fd34a_0 # # # # Downloading and Extracting Packages: ...working... done # Preparing transaction: ...working... done # Verifying transaction: ...working... done # Executing transaction: ...working... done # # # # To activate this environment, use # # # # $ conda activate ONTraC # # # # To deactivate an active environment, use # # # # $ conda deactivate # # Collecting git+https://github.com/gyuanlab/ONTraC.git@V2 # Cloning https://github.com/gyuanlab/ONTraC.git (to revision V2) to /private/var/ folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/pip-req-build-g2zbeni3 # Running command git clone --filter=blob:none --quiet https://github.com/gyuanlab/ ONTraC.git /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-req-build-g2zbeni3 # Running command git checkout -b V2 --track origin/V2 # Resolved https://github.com/gyuanlab/ONTraC.git to commit c2cf2355da9fd5dd580ad3d9d15321f0e3a92b9d # Installing build dependencies: started # Switched to a new branch 'V2' # branch 'V2' set up to track 'origin/V2'. # Installing build dependencies: finished with status 'done' # Getting requirements to build wheel: started # Getting requirements to build wheel: finished with status 'done' # Preparing metadata (pyproject.toml): started # Preparing metadata (pyproject.toml): finished with status 'done' # Collecting pyyaml==6.0.1 (from ONTraC==2.0a9.dev7) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.1 kB) # Collecting pandas==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (19 kB) # Collecting torch==2.2.1 (from ONTraC==2.0a9.dev7) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl.metadata (25 kB) # Collecting torch-geometric==2.5.0 (from ONTraC==2.0a9.dev7) # Using cached torch_geometric-2.5.0-py3-none-any.whl.metadata (64 kB) # Collecting umap-learn==0.5.6 (from ONTraC==2.0a9.dev7) # Using cached umap_learn-0.5.6-py3-none-any.whl.metadata (21 kB) # Collecting harmonypy==0.0.10 (from ONTraC==2.0a9.dev7) # Using cached harmonypy-0.0.10-py3-none-any.whl.metadata (3.9 kB) # Collecting leidenalg==0.10.2 (from ONTraC==2.0a9.dev7) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl.metadata (10 kB) # Collecting session-info (from ONTraC==2.0a9.dev7) # Using cached session_info-1.0.0-py3-none-any.whl # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Downloading numpy-2.0.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (60 kB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 60.9/60.9 kB 1.7 MB/s eta 0:00:00 # Collecting scikit-learn (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl.metadata (12 kB) # Collecting scipy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl.metadata (60 kB) # Collecting igraph<0.12,>=0.10.0 (from leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl.metadata (3.9 kB) # Collecting numpy (from harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (114 kB) # Collecting python-dateutil>=2.8.2 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl.metadata (8.4 kB) # Collecting pytz>=2020.1 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached pytz-2024.1-py2.py3-none-any.whl.metadata (22 kB) # Collecting tzdata>=2022.7 (from pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached tzdata-2024.1-py2.py3-none-any.whl.metadata (1.4 kB) # Collecting filelock (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached filelock-3.15.4-py3-none-any.whl.metadata (2.9 kB) # Collecting typing-extensions>=4.8.0 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached typing_extensions-4.12.2-py3-none-any.whl.metadata (3.0 kB) # Collecting sympy (from torch==2.2.1->ONTraC==2.0a9.dev7) # Downloading sympy-1.13.1-py3-none-any.whl.metadata (12 kB) # Collecting networkx (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached networkx-3.3-py3-none-any.whl.metadata (5.1 kB) # Collecting jinja2 (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached jinja2-3.1.4-py3-none-any.whl.metadata (2.6 kB) # Collecting fsspec (from torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached fsspec-2024.6.1-py3-none-any.whl.metadata (11 kB) # Collecting tqdm (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached tqdm-4.66.4-py3-none-any.whl.metadata (57 kB) # Collecting aiohttp (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (7.5 kB) # Collecting requests (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached requests-2.32.3-py3-none-any.whl.metadata (4.6 kB) # Collecting pyparsing (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached pyparsing-3.1.2-py3-none-any.whl.metadata (5.1 kB) # Collecting psutil>=5.8.0 (from torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl.metadata (21 kB) # Collecting numba>=0.51.2 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (2.7 kB) # Collecting pynndescent>=0.5 (from umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached pynndescent-0.5.13-py3-none-any.whl.metadata (6.8 kB) # Collecting stdlib-list (from session-info->ONTraC==2.0a9.dev7) # Using cached stdlib_list-0.10.0-py3-none-any.whl.metadata (3.3 kB) # Collecting texttable>=1.6.2 (from igraph< 0.12,>=0.10.0->leidenalg==0.10.2->ONTraC==2.0a9.dev7) # Using cached texttable-1.7.0-py2.py3-none-any.whl.metadata (9.8 kB) # Collecting llvmlite<0.44,>=0.43.0dev0 (from numba>=0.51.2->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.8 kB) # Collecting joblib>=0.11 (from pynndescent>=0.5->umap-learn==0.5.6->ONTraC==2.0a9.dev7) # Using cached joblib-1.4.2-py3-none-any.whl.metadata (5.4 kB) # Collecting six>=1.5 (from python-dateutil>=2.8.2->pandas==2.2.1->ONTraC==2.0a9.dev7) # Using cached six-1.16.0-py2.py3-none-any.whl.metadata (1.8 kB) # Collecting threadpoolctl>=3.1.0 (from scikit-learn->harmonypy==0.0.10->ONTraC==2.0a9.dev7) # Using cached threadpoolctl-3.5.0-py3-none-any.whl.metadata (13 kB) # Collecting aiosignal>=1.1.2 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached aiosignal-1.3.1-py3-none-any.whl.metadata (4.0 kB) # Collecting attrs>=17.3.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached attrs-23.2.0-py3-none-any.whl.metadata (9.5 kB) # Collecting frozenlist>=1.1.1 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl.metadata (12 kB) # Collecting multidict<7.0,>=4.5 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl.metadata (4.2 kB) # Collecting yarl<2.0,>=1.0 (from aiohttp->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl.metadata (31 kB) # Collecting MarkupSafe>=2.0 (from jinja2->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl.metadata (3.0 kB) # Collecting charset-normalizer<4,>=2 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl.metadata ( 33 kB) # Collecting idna<4,>=2.5 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached idna-3.7-py3-none-any.whl.metadata (9.9 kB) # Collecting urllib3<3,>=1.21.1 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached urllib3-2.2.2-py3-none-any.whl.metadata (6.4 kB) # Collecting certifi>=2017.4.17 (from requests->torch-geometric==2.5.0->ONTraC==2.0a9.dev7) # Using cached certifi-2024.7.4-py3-none-any.whl.metadata (2.2 kB) # Collecting mpmath<1.4,>=1.1.0 (from sympy->torch==2.2.1->ONTraC==2.0a9.dev7) # Using cached mpmath-1.3.0-py3-none-any.whl.metadata (8.6 kB) # Using cached harmonypy-0.0.10-py3-none-any.whl (20 kB) # Using cached leidenalg-0.10.2-cp38-abi3-macosx_11_0_arm64.whl (1.4 MB) # Using cached pandas-2.2.1-cp311-cp311-macosx_11_0_arm64.whl (11.3 MB) # Using cached PyYAML-6.0.1-cp311-cp311-macosx_11_0_arm64.whl (167 kB) # Using cached torch-2.2.1-cp311-none-macosx_11_0_arm64.whl (59.7 MB) # Using cached torch_geometric-2.5.0-py3-none-any.whl (1.1 MB) # Using cached umap_learn-0.5.6-py3-none-any.whl (85 kB) # Using cached igraph-0.11.6-cp39-abi3-macosx_11_0_arm64.whl (1.8 MB) # Using cached numba-0.60.0-cp311-cp311-macosx_11_0_arm64.whl (2.6 MB) # Using cached numpy-1.26.4-cp311-cp311-macosx_11_0_arm64.whl (14.0 MB) # Using cached psutil-6.0.0-cp38-abi3-macosx_11_0_arm64.whl (251 kB) # Using cached pynndescent-0.5.13-py3-none-any.whl (56 kB) # Using cached python_dateutil-2.9.0.post0-py2.py3-none-any.whl (229 kB) # Using cached pytz-2024.1-py2.py3-none-any.whl (505 kB) # Using cached scikit_learn-1.5.1-cp311-cp311-macosx_12_0_arm64.whl (11.0 MB) # Using cached scipy-1.14.0-cp311-cp311-macosx_12_0_arm64.whl (29.9 MB) # Using cached typing_extensions-4.12.2-py3-none-any.whl (37 kB) # Using cached tzdata-2024.1-py2.py3-none-any.whl (345 kB) # Using cached aiohttp-3.9.5-cp311-cp311-macosx_11_0_arm64.whl (390 kB) # Using cached filelock-3.15.4-py3-none-any.whl (16 kB) # Using cached fsspec-2024.6.1-py3-none-any.whl (177 kB) # Using cached jinja2-3.1.4-py3-none-any.whl (133 kB) # Using cached networkx-3.3-py3-none-any.whl (1.7 MB) # Using cached pyparsing-3.1.2-py3-none-any.whl (103 kB) # Using cached requests-2.32.3-py3-none-any.whl (64 kB) # Using cached stdlib_list-0.10.0-py3-none-any.whl (79 kB) # Downloading sympy-1.13.1-py3-none-any.whl (6.2 MB) # ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.2/6.2 MB 9.3 MB/s eta 0:00:00 # Using cached tqdm-4.66.4-py3-none-any.whl (78 kB) # Using cached aiosignal-1.3.1-py3-none-any.whl (7.6 kB) # Using cached attrs-23.2.0-py3-none-any.whl (60 kB) # Using cached certifi-2024.7.4-py3-none-any.whl (162 kB) # Using cached charset_normalizer-3.3.2-cp311-cp311-macosx_11_0_arm64.whl (118 kB) # Using cached frozenlist-1.4.1-cp311-cp311-macosx_11_0_arm64.whl (53 kB) # Using cached idna-3.7-py3-none-any.whl (66 kB) # Using cached joblib-1.4.2-py3-none-any.whl (301 kB) # Using cached llvmlite-0.43.0-cp311-cp311-macosx_11_0_arm64.whl (28.8 MB) # Using cached MarkupSafe-2.1.5-cp311-cp311-macosx_10_9_universal2.whl (18 kB) # Using cached mpmath-1.3.0-py3-none-any.whl (536 kB) # Using cached multidict-6.0.5-cp311-cp311-macosx_11_0_arm64.whl (30 kB) # Using cached six-1.16.0-py2.py3-none-any.whl (11 kB) # Using cached texttable-1.7.0-py2.py3-none-any.whl (10 kB) # Using cached threadpoolctl-3.5.0-py3-none-any.whl (18 kB) # Using cached urllib3-2.2.2-py3-none-any.whl (121 kB) # Using cached yarl-1.9.4-cp311-cp311-macosx_11_0_arm64.whl (81 kB) # Building wheels for collected packages: ONTraC # Building wheel for ONTraC (pyproject.toml): started # Building wheel for ONTraC (pyproject.toml): finished with status 'done' # Created wheel for ONTraC: filename=ONTraC-2.0a9.dev7-py3-none-any.whl size=56945 sha256=cc16ceec51ea66cf0036a568db4e43d052c2d41747e31f99c17fc4665d713179 # Stored in directory: /private/var/folders/4z/75w3m8r57jj3bkbbyx6shx7c0000gn/T/ pip-ephem-wheel-cache-7kgwlhji/wheels/88/64/83/ e8e4dfd8000c8e81e9724db9f092231791d3bcb083502fe37a # Successfully built ONTraC # Installing collected packages: texttable, pytz, mpmath, urllib3, tzdata, typing-extensions, tqdm, threadpoolctl, sympy, stdlib-list, six, pyyaml, pyparsing, psutil, numpy, networkx, multidict, MarkupSafe, llvmlite, joblib, igraph, idna, fsspec, frozenlist, filelock, charset-normalizer, certifi, attrs, yarl, session-info, scipy, requests, python-dateutil, numba, leidenalg, jinja2, aiosignal, torch, scikit-learn, pandas, aiohttp, torch-geometric, pynndescent, harmonypy, umap-learn, ONTraC # Successfully installed MarkupSafe-2.1.5 ONTraC-2.0a9.dev7 aiohttp-3.9.5 aiosignal-1.3.1 attrs-23.2.0 certifi-2024.7.4 charset-normalizer-3.3.2 filelock-3.15.4 frozenlist-1.4.1 fsspec-2024.6.1 harmonypy-0.0.10 idna-3.7 igraph-0.11.6 jinja2-3.1.4 joblib-1.4.2 leidenalg-0.10.2 llvmlite-0.43.0 mpmath-1.3.0 multidict-6.0.5 networkx-3.3 numba-0.60.0 numpy-1.26.4 pandas-2.2.1 psutil-6.0.0 pynndescent-0.5.13 pyparsing-3.1.2 python-dateutil-2.9.0.post0 pytz-2024.1 pyyaml-6.0.1 requests-2.32.3 scikit-learn-1.5.1 scipy-1.14.0 session-info-1.0.0 six-1.16.0 stdlib-list-0.10.0 sympy-1.13.1 texttable-1.7.0 threadpoolctl-3.5.0 torch-2.2.1 torch-geometric-2.5.0 tqdm-4.66.4 typing-extensions-4.12.2 tzdata-2024.1 umap-learn-0.5.6 urllib3-2.2.2 yarl-1.9.4 16.1.2.2 Running ONTraC source ~/.bash_profile conda activate ONTraC_v2_py311 ONTraC -d data/ONTraC_dataset_input.csv --preprocessing-dir data/preprocessing_dir --GNN-dir data/GNN_dir --NTScore-dir data/NTScore_dir --device cuda --epochs 1000 -s 42 --patience 100 --min-delta 0.001 --min-epochs 50 --lr 0.03 --hidden-feats 4 -k 6 --modularity-loss-weight 1 --regularization-loss-weight 0.1 --purity-loss-weight 100 --beta 0.3 --equal-space 2>&1 | tee data/merfish_subset.log # ################################################################################## # # ▄▄█▀▀██ ▀█▄ ▀█▀ █▀▀██▀▀█ ▄▄█▀▀▀▄█ # ▄█▀ ██ █▀█ █ ██ ▄▄▄ ▄▄ ▄▄▄▄ ▄█▀ ▀ # ██ ██ █ ▀█▄ █ ██ ██▀ ▀▀ ▀▀ ▄██ ██ # ▀█▄ ██ █ ███ ██ ██ ▄█▀ ██ ▀█▄ ▄ # ▀▀█▄▄▄█▀ ▄█▄ ▀█ ▄██▄ ▄██▄ ▀█▄▄▀█▀ ▀▀█▄▄▄▄▀ # # version: 2.0a11 # # ################################################################################## # 11:33:23 --- WARNING: The directory (data/preprocessing_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/GNN_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: The directory (data/NTScore_dir) you given already exists. It will be overwritten. # 11:33:23 --- WARNING: CUDA is not available, use CPU instead. # 11:33:23 --- INFO: ------------------ RUN params memo ------------------ # 11:33:23 --- INFO: -------- I/O options ------- # 11:33:23 --- INFO: meta data file: data/ONTraC_dataset_input.csv # 11:33:23 --- INFO: preprocessing output directory: data/preprocessing_dir # 11:33:23 --- INFO: GNN output directory: data/GNN_dir # 11:33:23 --- INFO: NTScore output directory: data/NTScore_dir # 11:33:23 --- INFO: -------- preprocessing options ------- # 11:33:23 --- INFO: resolution: 10.0 # 11:33:23 --- INFO: -------- niche net constr options ------- # 11:33:23 --- INFO: n_cpu: 4 # 11:33:23 --- INFO: n_neighbors: 50 # 11:33:23 --- INFO: n_local: 20 # 11:33:23 --- INFO: embedding_adjust: False # 11:33:23 --- INFO: sigma: 1 # 11:33:23 --- INFO: -------- train options ------- # 11:33:23 --- INFO: device: cpu # 11:33:23 --- INFO: epochs: 1000 # 11:33:23 --- INFO: batch_size: 0 # 11:33:23 --- INFO: patience: 100 # 11:33:23 --- INFO: min_delta: 0.001 # 11:33:23 --- INFO: min_epochs: 50 # 11:33:23 --- INFO: seed: 42 # 11:33:23 --- INFO: lr: 0.03 # 11:33:23 --- INFO: hidden_feats: 4 # 11:33:23 --- INFO: k: 6 # 11:33:23 --- INFO: modularity_loss_weight: 1.0 # 11:33:23 --- INFO: purity_loss_weight: 100.0 # 11:33:23 --- INFO: regularization_loss_weight: 0.1 # 11:33:23 --- INFO: beta: 0.3 # 11:33:23 --- INFO: ---------------- Niche trajectory options ---------------- # 11:33:23 --- INFO: Equally spaced niche cluster scores: True # 11:33:23 --- INFO: --------------- RUN params memo end ----------------- # 11:33:23 --- INFO: ------------- Niche network construct --------------- # 11:33:23 --- INFO: Constructing niche network for sample: mouse2_slice229. # 11:33:26 --- INFO: Constructing niche network for sample: mouse2_slice300. # 11:33:28 --- INFO: Generating samples.yaml file. # 11:33:28 --- INFO: ------------ Niche network construct end ------------ # 11:33:28 --- INFO: ------------------------ GNN ------------------------ # 11:33:28 --- INFO: Loading dataset. # 11:33:28 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:33:28 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:33:28 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:33:28 --- INFO: epoch: 1, batch: 1, loss: 23.001523971557617, modularity_loss: -0.0023897262290120125, purity_loss: 22.934659957885742, regularization_loss: 0.06925322115421295 # 11:33:29 --- INFO: epoch: 2, batch: 1, loss: 22.768497467041016, modularity_loss: -0.005293438211083412, purity_loss: 22.704635620117188, regularization_loss: 0.06915470957756042 # 11:33:29 --- INFO: epoch: 3, batch: 1, loss: 22.37835121154785, modularity_loss: -0.010112201794981956, purity_loss: 22.319320678710938, regularization_loss: 0.06914201378822327 # 11:33:29 --- INFO: epoch: 4, batch: 1, loss: 21.823326110839844, modularity_loss: -0.016986437141895294, purity_loss: 21.77100944519043, regularization_loss: 0.06930312514305115 # 11:33:30 --- INFO: epoch: 5, batch: 1, loss: 21.139827728271484, modularity_loss: -0.025495745241642, purity_loss: 21.095653533935547, regularization_loss: 0.0696694627404213 # 11:33:30 --- INFO: epoch: 6, batch: 1, loss: 20.39137077331543, modularity_loss: -0.03495821729302406, purity_loss: 20.356155395507812, regularization_loss: 0.0701739713549614 # 11:33:30 --- INFO: epoch: 7, batch: 1, loss: 19.641571044921875, modularity_loss: -0.045391954481601715, purity_loss: 19.616260528564453, regularization_loss: 0.07070285081863403 # 11:33:31 --- INFO: epoch: 8, batch: 1, loss: 18.925037384033203, modularity_loss: -0.05757240578532219, purity_loss: 18.911428451538086, regularization_loss: 0.0711822658777237 # 11:33:31 --- INFO: epoch: 9, batch: 1, loss: 18.263259887695312, modularity_loss: -0.0717809870839119, purity_loss: 18.263416290283203, regularization_loss: 0.07162471860647202 # 11:33:31 --- INFO: epoch: 10, batch: 1, loss: 17.685840606689453, modularity_loss: -0.08731567114591599, purity_loss: 17.701066970825195, regularization_loss: 0.07208941131830215 # 11:33:31 --- INFO: epoch: 11, batch: 1, loss: 17.217161178588867, modularity_loss: -0.10291540622711182, purity_loss: 17.247447967529297, regularization_loss: 0.07262824475765228 # 11:33:32 --- INFO: epoch: 12, batch: 1, loss: 16.85984230041504, modularity_loss: -0.11742901802062988, purity_loss: 16.904020309448242, regularization_loss: 0.07325152307748795 # 11:33:32 --- INFO: epoch: 13, batch: 1, loss: 16.59726905822754, modularity_loss: -0.13028934597969055, purity_loss: 16.653614044189453, regularization_loss: 0.07394523918628693 # 11:33:32 --- INFO: epoch: 14, batch: 1, loss: 16.409076690673828, modularity_loss: -0.14140018820762634, purity_loss: 16.47577476501465, regularization_loss: 0.07470250874757767 # 11:33:33 --- INFO: epoch: 15, batch: 1, loss: 16.27598762512207, modularity_loss: -0.15096718072891235, purity_loss: 16.351421356201172, regularization_loss: 0.07553426176309586 # 11:33:33 --- INFO: epoch: 16, batch: 1, loss: 16.18242645263672, modularity_loss: -0.15935572981834412, purity_loss: 16.26532745361328, regularization_loss: 0.07645474374294281 # 11:33:33 --- INFO: epoch: 17, batch: 1, loss: 16.117395401000977, modularity_loss: -0.16692203283309937, purity_loss: 16.20685577392578, regularization_loss: 0.07746140658855438 # 11:33:33 --- INFO: epoch: 18, batch: 1, loss: 16.072946548461914, modularity_loss: -0.17391526699066162, purity_loss: 16.168333053588867, regularization_loss: 0.07852821052074432 # 11:33:34 --- INFO: epoch: 19, batch: 1, loss: 16.042552947998047, modularity_loss: -0.18049469590187073, purity_loss: 16.143430709838867, regularization_loss: 0.07961639761924744 # 11:33:34 --- INFO: epoch: 20, batch: 1, loss: 16.020952224731445, modularity_loss: -0.18679285049438477, purity_loss: 16.12704849243164, regularization_loss: 0.08069746196269989 # 11:33:34 --- INFO: epoch: 21, batch: 1, loss: 16.004188537597656, modularity_loss: -0.19300395250320435, purity_loss: 16.11542510986328, regularization_loss: 0.08176703751087189 # 11:33:35 --- INFO: epoch: 22, batch: 1, loss: 15.989411354064941, modularity_loss: -0.19940897822380066, purity_loss: 16.105968475341797, regularization_loss: 0.0828515961766243 # 11:33:35 --- INFO: epoch: 23, batch: 1, loss: 15.974446296691895, modularity_loss: -0.20637741684913635, purity_loss: 16.096820831298828, regularization_loss: 0.08400280028581619 # 11:33:35 --- INFO: epoch: 24, batch: 1, loss: 15.957433700561523, modularity_loss: -0.2143288552761078, purity_loss: 16.08647346496582, regularization_loss: 0.08528861403465271 # 11:33:35 --- INFO: epoch: 25, batch: 1, loss: 15.936773300170898, modularity_loss: -0.2236764132976532, purity_loss: 16.073673248291016, regularization_loss: 0.08677610009908676 # 11:33:36 --- INFO: epoch: 26, batch: 1, loss: 15.911301612854004, modularity_loss: -0.23471668362617493, purity_loss: 16.057510375976562, regularization_loss: 0.08850813657045364 # 11:33:36 --- INFO: epoch: 27, batch: 1, loss: 15.880578994750977, modularity_loss: -0.24747242033481598, purity_loss: 16.037572860717773, regularization_loss: 0.09047902375459671 # 11:33:36 --- INFO: epoch: 28, batch: 1, loss: 15.845392227172852, modularity_loss: -0.26156920194625854, purity_loss: 16.014341354370117, regularization_loss: 0.09261994063854218 # 11:33:37 --- INFO: epoch: 29, batch: 1, loss: 15.807584762573242, modularity_loss: -0.27627527713775635, purity_loss: 15.989053726196289, regularization_loss: 0.09480661898851395 # 11:33:37 --- INFO: epoch: 30, batch: 1, loss: 15.769493103027344, modularity_loss: -0.2906855046749115, purity_loss: 15.963276863098145, regularization_loss: 0.09690161794424057 # 11:33:37 --- INFO: epoch: 31, batch: 1, loss: 15.733196258544922, modularity_loss: -0.3040352165699005, purity_loss: 15.938435554504395, regularization_loss: 0.09879627078771591 # 11:33:37 --- INFO: epoch: 32, batch: 1, loss: 15.699841499328613, modularity_loss: -0.31587138772010803, purity_loss: 15.915274620056152, regularization_loss: 0.10043793171644211 # 11:33:38 --- INFO: epoch: 33, batch: 1, loss: 15.669454574584961, modularity_loss: -0.32608187198638916, purity_loss: 15.89371109008789, regularization_loss: 0.1018255203962326 # 11:33:38 --- INFO: epoch: 34, batch: 1, loss: 15.641484260559082, modularity_loss: -0.33480289578437805, purity_loss: 15.873299598693848, regularization_loss: 0.10298792272806168 # 11:33:38 --- INFO: epoch: 35, batch: 1, loss: 15.614999771118164, modularity_loss: -0.34228256344795227, purity_loss: 15.853317260742188, regularization_loss: 0.10396471619606018 # 11:33:39 --- INFO: epoch: 36, batch: 1, loss: 15.589118003845215, modularity_loss: -0.3488248586654663, purity_loss: 15.833154678344727, regularization_loss: 0.10478848218917847 # 11:33:39 --- INFO: epoch: 37, batch: 1, loss: 15.56322193145752, modularity_loss: -0.35469937324523926, purity_loss: 15.812437057495117, regularization_loss: 0.1054840087890625 # 11:33:39 --- INFO: epoch: 38, batch: 1, loss: 15.536772727966309, modularity_loss: -0.3601019084453583, purity_loss: 15.790804862976074, regularization_loss: 0.10606933385133743 # 11:33:39 --- INFO: epoch: 39, batch: 1, loss: 15.508807182312012, modularity_loss: -0.36518266797065735, purity_loss: 15.767438888549805, regularization_loss: 0.10655105113983154 # 11:33:40 --- INFO: epoch: 40, batch: 1, loss: 15.478368759155273, modularity_loss: -0.3700413703918457, purity_loss: 15.741480827331543, regularization_loss: 0.10692881792783737 # 11:33:40 --- INFO: epoch: 41, batch: 1, loss: 15.44459342956543, modularity_loss: -0.37471112608909607, purity_loss: 15.712104797363281, regularization_loss: 0.10719987750053406 # 11:33:40 --- INFO: epoch: 42, batch: 1, loss: 15.40697956085205, modularity_loss: -0.37914952635765076, purity_loss: 15.678765296936035, regularization_loss: 0.10736335813999176 # 11:33:41 --- INFO: epoch: 43, batch: 1, loss: 15.364958763122559, modularity_loss: -0.38326019048690796, purity_loss: 15.640804290771484, regularization_loss: 0.10741502791643143 # 11:33:41 --- INFO: epoch: 44, batch: 1, loss: 15.317839622497559, modularity_loss: -0.38691914081573486, purity_loss: 15.597410202026367, regularization_loss: 0.10734901577234268 # 11:33:41 --- INFO: epoch: 45, batch: 1, loss: 15.265110969543457, modularity_loss: -0.38995009660720825, purity_loss: 15.547901153564453, regularization_loss: 0.10716016590595245 # 11:33:41 --- INFO: epoch: 46, batch: 1, loss: 15.20550537109375, modularity_loss: -0.3921312093734741, purity_loss: 15.490798950195312, regularization_loss: 0.10683741420507431 # 11:33:42 --- INFO: epoch: 47, batch: 1, loss: 15.136863708496094, modularity_loss: -0.39331942796707153, purity_loss: 15.423828125, regularization_loss: 0.10635543614625931 # 11:33:42 --- INFO: epoch: 48, batch: 1, loss: 15.056995391845703, modularity_loss: -0.3934229612350464, purity_loss: 15.34473705291748, regularization_loss: 0.10568106919527054 # 11:33:42 --- INFO: epoch: 49, batch: 1, loss: 14.964367866516113, modularity_loss: -0.392358660697937, purity_loss: 15.251948356628418, regularization_loss: 0.10477815568447113 # 11:33:43 --- INFO: epoch: 50, batch: 1, loss: 14.857380867004395, modularity_loss: -0.39002490043640137, purity_loss: 15.143804550170898, regularization_loss: 0.10360138863325119 # 11:33:43 --- INFO: epoch: 51, batch: 1, loss: 14.736187934875488, modularity_loss: -0.3862961530685425, purity_loss: 15.02037525177002, regularization_loss: 0.10210883617401123 # 11:33:43 --- INFO: epoch: 52, batch: 1, loss: 14.603105545043945, modularity_loss: -0.38095587491989136, purity_loss: 14.883785247802734, regularization_loss: 0.10027574747800827 # 11:33:43 --- INFO: epoch: 53, batch: 1, loss: 14.458536148071289, modularity_loss: -0.3737679719924927, purity_loss: 14.734207153320312, regularization_loss: 0.09809701144695282 # 11:33:44 --- INFO: epoch: 54, batch: 1, loss: 14.297077178955078, modularity_loss: -0.36470723152160645, purity_loss: 14.566164016723633, regularization_loss: 0.09562009572982788 # 11:33:44 --- INFO: epoch: 55, batch: 1, loss: 14.101924896240234, modularity_loss: -0.35435616970062256, purity_loss: 14.36331558227539, regularization_loss: 0.09296524524688721 # 11:33:44 --- INFO: epoch: 56, batch: 1, loss: 13.850761413574219, modularity_loss: -0.3440517783164978, purity_loss: 14.104469299316406, regularization_loss: 0.09034416824579239 # 11:33:45 --- INFO: epoch: 57, batch: 1, loss: 13.542003631591797, modularity_loss: -0.33538782596588135, purity_loss: 13.789325714111328, regularization_loss: 0.08806595206260681 # 11:33:45 --- INFO: epoch: 58, batch: 1, loss: 13.19692611694336, modularity_loss: -0.3292675316333771, purity_loss: 13.43971061706543, regularization_loss: 0.08648291230201721 # 11:33:45 --- INFO: epoch: 59, batch: 1, loss: 12.85534954071045, modularity_loss: -0.3257511854171753, purity_loss: 13.095155715942383, regularization_loss: 0.08594459295272827 # 11:33:45 --- INFO: epoch: 60, batch: 1, loss: 12.5657958984375, modularity_loss: -0.32381701469421387, purity_loss: 12.803165435791016, regularization_loss: 0.08644744753837585 # 11:33:46 --- INFO: epoch: 61, batch: 1, loss: 12.347742080688477, modularity_loss: -0.3218806982040405, purity_loss: 12.58204460144043, regularization_loss: 0.08757861703634262 # 11:33:46 --- INFO: epoch: 62, batch: 1, loss: 12.205147743225098, modularity_loss: -0.31888464093208313, purity_loss: 12.435131072998047, regularization_loss: 0.08890123665332794 # 11:33:46 --- INFO: epoch: 63, batch: 1, loss: 12.128698348999023, modularity_loss: -0.31569480895996094, purity_loss: 12.35433292388916, regularization_loss: 0.09006011486053467 # 11:33:47 --- INFO: epoch: 64, batch: 1, loss: 12.073147773742676, modularity_loss: -0.3147158622741699, purity_loss: 12.297168731689453, regularization_loss: 0.09069512784481049 # 11:33:47 --- INFO: epoch: 65, batch: 1, loss: 11.988947868347168, modularity_loss: -0.3177931010723114, purity_loss: 12.216124534606934, regularization_loss: 0.09061658382415771 # 11:33:47 --- INFO: epoch: 66, batch: 1, loss: 11.866996765136719, modularity_loss: -0.32488876581192017, purity_loss: 12.101926803588867, regularization_loss: 0.08995886892080307 # 11:33:48 --- INFO: epoch: 67, batch: 1, loss: 11.727216720581055, modularity_loss: -0.33459246158599854, purity_loss: 11.972763061523438, regularization_loss: 0.0890461802482605 # 11:33:48 --- INFO: epoch: 68, batch: 1, loss: 11.589557647705078, modularity_loss: -0.3454422354698181, purity_loss: 11.846831321716309, regularization_loss: 0.08816889673471451 # 11:33:48 --- INFO: epoch: 69, batch: 1, loss: 11.461546897888184, modularity_loss: -0.3565739095211029, purity_loss: 11.730629920959473, regularization_loss: 0.0874905213713646 # 11:33:48 --- INFO: epoch: 70, batch: 1, loss: 11.341334342956543, modularity_loss: -0.3676247000694275, purity_loss: 11.621889114379883, regularization_loss: 0.08707026392221451 # 11:33:49 --- INFO: epoch: 71, batch: 1, loss: 11.220848083496094, modularity_loss: -0.37854552268981934, purity_loss: 11.512494087219238, regularization_loss: 0.08689992129802704 # 11:33:49 --- INFO: epoch: 72, batch: 1, loss: 11.089977264404297, modularity_loss: -0.38959217071533203, purity_loss: 11.392634391784668, regularization_loss: 0.08693493902683258 # 11:33:49 --- INFO: epoch: 73, batch: 1, loss: 10.936225891113281, modularity_loss: -0.40123844146728516, purity_loss: 11.250344276428223, regularization_loss: 0.08711998164653778 # 11:33:50 --- INFO: epoch: 74, batch: 1, loss: 10.774359703063965, modularity_loss: -0.412983775138855, purity_loss: 11.099967956542969, regularization_loss: 0.0873752236366272 # 11:33:50 --- INFO: epoch: 75, batch: 1, loss: 10.597075462341309, modularity_loss: -0.4235861301422119, purity_loss: 10.933009147644043, regularization_loss: 0.08765210211277008 # 11:33:50 --- INFO: epoch: 76, batch: 1, loss: 10.376021385192871, modularity_loss: -0.43360933661460876, purity_loss: 10.721734046936035, regularization_loss: 0.08789674937725067 # 11:33:51 --- INFO: epoch: 77, batch: 1, loss: 10.101761817932129, modularity_loss: -0.44318681955337524, purity_loss: 10.456952095031738, regularization_loss: 0.08799697458744049 # 11:33:51 --- INFO: epoch: 78, batch: 1, loss: 9.784876823425293, modularity_loss: -0.4515224099159241, purity_loss: 10.148638725280762, regularization_loss: 0.08776087313890457 # 11:33:51 --- INFO: epoch: 79, batch: 1, loss: 9.458683013916016, modularity_loss: -0.4569146931171417, purity_loss: 9.828640937805176, regularization_loss: 0.08695651590824127 # 11:33:52 --- INFO: epoch: 80, batch: 1, loss: 9.178531646728516, modularity_loss: -0.45679569244384766, purity_loss: 9.549927711486816, regularization_loss: 0.08539916574954987 # 11:33:52 --- INFO: epoch: 81, batch: 1, loss: 9.000457763671875, modularity_loss: -0.4497307538986206, purity_loss: 9.366915702819824, regularization_loss: 0.08327241241931915 # 11:33:52 --- INFO: epoch: 82, batch: 1, loss: 8.898308753967285, modularity_loss: -0.4418599605560303, purity_loss: 9.258613586425781, regularization_loss: 0.08155520260334015 # 11:33:52 --- INFO: epoch: 83, batch: 1, loss: 8.760506629943848, modularity_loss: -0.44438159465789795, purity_loss: 9.123655319213867, regularization_loss: 0.08123327046632767 # 11:33:53 --- INFO: epoch: 84, batch: 1, loss: 8.54394245147705, modularity_loss: -0.45904988050460815, purity_loss: 8.920684814453125, regularization_loss: 0.08230743557214737 # 11:33:53 --- INFO: epoch: 85, batch: 1, loss: 8.314882278442383, modularity_loss: -0.4775947332382202, purity_loss: 8.708457946777344, regularization_loss: 0.08401863276958466 # 11:33:53 --- INFO: epoch: 86, batch: 1, loss: 8.135581970214844, modularity_loss: -0.492197185754776, purity_loss: 8.542383193969727, regularization_loss: 0.08539611101150513 # 11:33:54 --- INFO: epoch: 87, batch: 1, loss: 7.994486331939697, modularity_loss: -0.5015395879745483, purity_loss: 8.410036087036133, regularization_loss: 0.08598977327346802 # 11:33:54 --- INFO: epoch: 88, batch: 1, loss: 7.859262943267822, modularity_loss: -0.5070834159851074, purity_loss: 8.280491828918457, regularization_loss: 0.08585438132286072 # 11:33:54 --- INFO: epoch: 89, batch: 1, loss: 7.714503765106201, modularity_loss: -0.5096077919006348, purity_loss: 8.138916969299316, regularization_loss: 0.08519440144300461 # 11:33:55 --- INFO: epoch: 90, batch: 1, loss: 7.556613445281982, modularity_loss: -0.5087662935256958, purity_loss: 7.981185436248779, regularization_loss: 0.08419422060251236 # 11:33:55 --- INFO: epoch: 91, batch: 1, loss: 7.387192249298096, modularity_loss: -0.5037617683410645, purity_loss: 7.807967662811279, regularization_loss: 0.08298640698194504 # 11:33:55 --- INFO: epoch: 92, batch: 1, loss: 7.229267597198486, modularity_loss: -0.4948621988296509, purity_loss: 7.642430782318115, regularization_loss: 0.08169892430305481 # 11:33:56 --- INFO: epoch: 93, batch: 1, loss: 7.137825012207031, modularity_loss: -0.48517730832099915, purity_loss: 7.542435169219971, regularization_loss: 0.08056731522083282 # 11:33:56 --- INFO: epoch: 94, batch: 1, loss: 7.1067352294921875, modularity_loss: -0.47995471954345703, purity_loss: 7.5068511962890625, regularization_loss: 0.079838827252388 # 11:33:56 --- INFO: epoch: 95, batch: 1, loss: 7.045711994171143, modularity_loss: -0.48180586099624634, purity_loss: 7.448028087615967, regularization_loss: 0.0794895812869072 # 11:33:57 --- INFO: epoch: 96, batch: 1, loss: 6.957595348358154, modularity_loss: -0.48858559131622314, purity_loss: 7.366804122924805, regularization_loss: 0.07937701046466827 # 11:33:57 --- INFO: epoch: 97, batch: 1, loss: 6.891050815582275, modularity_loss: -0.49676376581192017, purity_loss: 7.308403968811035, regularization_loss: 0.0794105976819992 # 11:33:57 --- INFO: epoch: 98, batch: 1, loss: 6.855169773101807, modularity_loss: -0.5040184855461121, purity_loss: 7.279667377471924, regularization_loss: 0.07952070236206055 # 11:33:57 --- INFO: epoch: 99, batch: 1, loss: 6.834170818328857, modularity_loss: -0.509428858757019, purity_loss: 7.263950347900391, regularization_loss: 0.07964947819709778 # 11:33:58 --- INFO: epoch: 100, batch: 1, loss: 6.814302921295166, modularity_loss: -0.5128939151763916, purity_loss: 7.247436046600342, regularization_loss: 0.07976070791482925 # 11:33:58 --- INFO: epoch: 101, batch: 1, loss: 6.791234493255615, modularity_loss: -0.5147401094436646, purity_loss: 7.226131916046143, regularization_loss: 0.07984252274036407 # 11:33:58 --- INFO: epoch: 102, batch: 1, loss: 6.767107963562012, modularity_loss: -0.5154187679290771, purity_loss: 7.202628135681152, regularization_loss: 0.07989867776632309 # 11:33:59 --- INFO: epoch: 103, batch: 1, loss: 6.745961666107178, modularity_loss: -0.5153516530990601, purity_loss: 7.1813764572143555, regularization_loss: 0.07993695139884949 # 11:33:59 --- INFO: epoch: 104, batch: 1, loss: 6.73048734664917, modularity_loss: -0.5148610472679138, purity_loss: 7.165385723114014, regularization_loss: 0.07996267825365067 # 11:33:59 --- INFO: epoch: 105, batch: 1, loss: 6.7206220626831055, modularity_loss: -0.5141782760620117, purity_loss: 7.154825210571289, regularization_loss: 0.07997535914182663 # 11:34:00 --- INFO: epoch: 106, batch: 1, loss: 6.713866233825684, modularity_loss: -0.5134555101394653, purity_loss: 7.147350788116455, regularization_loss: 0.07997094839811325 # 11:34:00 --- INFO: epoch: 107, batch: 1, loss: 6.707263469696045, modularity_loss: -0.5127870440483093, purity_loss: 7.140103816986084, regularization_loss: 0.07994650304317474 # 11:34:00 --- INFO: epoch: 108, batch: 1, loss: 6.698901176452637, modularity_loss: -0.5122053027153015, purity_loss: 7.13120698928833, regularization_loss: 0.07989972829818726 # 11:34:01 --- INFO: epoch: 109, batch: 1, loss: 6.688510417938232, modularity_loss: -0.5117073059082031, purity_loss: 7.120386600494385, regularization_loss: 0.07983095198869705 # 11:34:01 --- INFO: epoch: 110, batch: 1, loss: 6.6772356033325195, modularity_loss: -0.5112631320953369, purity_loss: 7.1087541580200195, regularization_loss: 0.07974453270435333 # 11:34:01 --- INFO: epoch: 111, batch: 1, loss: 6.666220188140869, modularity_loss: -0.510830283164978, purity_loss: 7.097405433654785, regularization_loss: 0.07964499294757843 # 11:34:01 --- INFO: epoch: 112, batch: 1, loss: 6.656651496887207, modularity_loss: -0.5103691816329956, purity_loss: 7.087483882904053, regularization_loss: 0.07953672111034393 # 11:34:02 --- INFO: epoch: 113, batch: 1, loss: 6.649173736572266, modularity_loss: -0.5098740458488464, purity_loss: 7.079622745513916, regularization_loss: 0.07942516356706619 # 11:34:02 --- INFO: epoch: 114, batch: 1, loss: 6.643716812133789, modularity_loss: -0.5093961358070374, purity_loss: 7.073794364929199, regularization_loss: 0.07931867986917496 # 11:34:02 --- INFO: epoch: 115, batch: 1, loss: 6.639582633972168, modularity_loss: -0.509020209312439, purity_loss: 7.0693769454956055, regularization_loss: 0.07922565191984177 # 11:34:03 --- INFO: epoch: 116, batch: 1, loss: 6.635583400726318, modularity_loss: -0.5088145732879639, purity_loss: 7.065243244171143, regularization_loss: 0.07915476709604263 # 11:34:03 --- INFO: epoch: 117, batch: 1, loss: 6.630502700805664, modularity_loss: -0.5087877511978149, purity_loss: 7.060179233551025, regularization_loss: 0.07911141961812973 # 11:34:03 --- INFO: epoch: 118, batch: 1, loss: 6.623572826385498, modularity_loss: -0.5089311599731445, purity_loss: 7.05340576171875, regularization_loss: 0.07909806072711945 # 11:34:04 --- INFO: epoch: 119, batch: 1, loss: 6.614688396453857, modularity_loss: -0.5092304944992065, purity_loss: 7.04480504989624, regularization_loss: 0.07911382615566254 # 11:34:04 --- INFO: epoch: 120, batch: 1, loss: 6.6042094230651855, modularity_loss: -0.5096694231033325, purity_loss: 7.034723281860352, regularization_loss: 0.07915548980236053 # 11:34:04 --- INFO: epoch: 121, batch: 1, loss: 6.592793941497803, modularity_loss: -0.5102277398109436, purity_loss: 7.0238037109375, regularization_loss: 0.07921795547008514 # 11:34:05 --- INFO: epoch: 122, batch: 1, loss: 6.581190586090088, modularity_loss: -0.5108745098114014, purity_loss: 7.012771129608154, regularization_loss: 0.07929384708404541 # 11:34:05 --- INFO: epoch: 123, batch: 1, loss: 6.56988000869751, modularity_loss: -0.5115731954574585, purity_loss: 7.002078533172607, regularization_loss: 0.07937465608119965 # 11:34:05 --- INFO: epoch: 124, batch: 1, loss: 6.55911111831665, modularity_loss: -0.5122793912887573, purity_loss: 6.991938591003418, regularization_loss: 0.07945197075605392 # 11:34:06 --- INFO: epoch: 125, batch: 1, loss: 6.548907279968262, modularity_loss: -0.5129407644271851, purity_loss: 6.9823317527771, regularization_loss: 0.07951626181602478 # 11:34:06 --- INFO: epoch: 126, batch: 1, loss: 6.539026260375977, modularity_loss: -0.513504147529602, purity_loss: 6.972971439361572, regularization_loss: 0.0795588493347168 # 11:34:06 --- INFO: epoch: 127, batch: 1, loss: 6.529170036315918, modularity_loss: -0.5139155387878418, purity_loss: 6.963512897491455, regularization_loss: 0.07957267761230469 # 11:34:06 --- INFO: epoch: 128, batch: 1, loss: 6.51899528503418, modularity_loss: -0.5141267776489258, purity_loss: 6.953570365905762, regularization_loss: 0.07955189049243927 # 11:34:07 --- INFO: epoch: 129, batch: 1, loss: 6.508218288421631, modularity_loss: -0.5141019225120544, purity_loss: 6.942827224731445, regularization_loss: 0.07949298620223999 # 11:34:07 --- INFO: epoch: 130, batch: 1, loss: 6.496604919433594, modularity_loss: -0.5138121247291565, purity_loss: 6.931023120880127, regularization_loss: 0.0793939158320427 # 11:34:07 --- INFO: epoch: 131, batch: 1, loss: 6.483977317810059, modularity_loss: -0.5132466554641724, purity_loss: 6.917969226837158, regularization_loss: 0.07925453782081604 # 11:34:08 --- INFO: epoch: 132, batch: 1, loss: 6.47015905380249, modularity_loss: -0.5124086141586304, purity_loss: 6.903491020202637, regularization_loss: 0.07907651364803314 # 11:34:08 --- INFO: epoch: 133, batch: 1, loss: 6.455018043518066, modularity_loss: -0.5113234519958496, purity_loss: 6.887477874755859, regularization_loss: 0.07886337488889694 # 11:34:08 --- INFO: epoch: 134, batch: 1, loss: 6.438426494598389, modularity_loss: -0.5100424289703369, purity_loss: 6.869848728179932, regularization_loss: 0.07862036675214767 # 11:34:08 --- INFO: epoch: 135, batch: 1, loss: 6.420292377471924, modularity_loss: -0.5086501240730286, purity_loss: 6.850588321685791, regularization_loss: 0.07835423201322556 # 11:34:09 --- INFO: epoch: 136, batch: 1, loss: 6.400551795959473, modularity_loss: -0.5072620511054993, purity_loss: 6.829740047454834, regularization_loss: 0.07807356864213943 # 11:34:09 --- INFO: epoch: 137, batch: 1, loss: 6.379120826721191, modularity_loss: -0.5060297250747681, purity_loss: 6.807362079620361, regularization_loss: 0.07778850197792053 # 11:34:09 --- INFO: epoch: 138, batch: 1, loss: 6.355955600738525, modularity_loss: -0.5051349401473999, purity_loss: 6.783579349517822, regularization_loss: 0.07751131057739258 # 11:34:10 --- INFO: epoch: 139, batch: 1, loss: 6.33098030090332, modularity_loss: -0.5047597289085388, purity_loss: 6.758484363555908, regularization_loss: 0.07725586742162704 # 11:34:10 --- INFO: epoch: 140, batch: 1, loss: 6.304221153259277, modularity_loss: -0.5050544738769531, purity_loss: 6.732240200042725, regularization_loss: 0.07703562825918198 # 11:34:10 --- INFO: epoch: 141, batch: 1, loss: 6.275861740112305, modularity_loss: -0.5061154961585999, purity_loss: 6.705114364624023, regularization_loss: 0.07686273753643036 # 11:34:11 --- INFO: epoch: 142, batch: 1, loss: 6.2462382316589355, modularity_loss: -0.5079628229141235, purity_loss: 6.677453994750977, regularization_loss: 0.0767468735575676 # 11:34:11 --- INFO: epoch: 143, batch: 1, loss: 6.215826034545898, modularity_loss: -0.510533332824707, purity_loss: 6.6496663093566895, regularization_loss: 0.07669306546449661 # 11:34:11 --- INFO: epoch: 144, batch: 1, loss: 6.185096740722656, modularity_loss: -0.5137086510658264, purity_loss: 6.622103691101074, regularization_loss: 0.0767018273472786 # 11:34:11 --- INFO: epoch: 145, batch: 1, loss: 6.154420852661133, modularity_loss: -0.517345666885376, purity_loss: 6.594995975494385, regularization_loss: 0.0767703503370285 # 11:34:12 --- INFO: epoch: 146, batch: 1, loss: 6.123962879180908, modularity_loss: -0.5213078856468201, purity_loss: 6.568376541137695, regularization_loss: 0.07689441740512848 # 11:34:12 --- INFO: epoch: 147, batch: 1, loss: 6.093730926513672, modularity_loss: -0.5254709124565125, purity_loss: 6.542133331298828, regularization_loss: 0.07706832140684128 # 11:34:12 --- INFO: epoch: 148, batch: 1, loss: 6.063621997833252, modularity_loss: -0.5297391414642334, purity_loss: 6.516074180603027, regularization_loss: 0.07728695869445801 # 11:34:13 --- INFO: epoch: 149, batch: 1, loss: 6.033446788787842, modularity_loss: -0.5340426564216614, purity_loss: 6.4899444580078125, regularization_loss: 0.07754483073949814 # 11:34:13 --- INFO: epoch: 150, batch: 1, loss: 6.002956867218018, modularity_loss: -0.538323163986206, purity_loss: 6.463444232940674, regularization_loss: 0.07783565670251846 # 11:34:13 --- INFO: epoch: 151, batch: 1, loss: 5.972036361694336, modularity_loss: -0.5425262451171875, purity_loss: 6.4364118576049805, regularization_loss: 0.07815083116292953 # 11:34:14 --- INFO: epoch: 152, batch: 1, loss: 5.940959453582764, modularity_loss: -0.5465943813323975, purity_loss: 6.4090752601623535, regularization_loss: 0.07847876101732254 # 11:34:14 --- INFO: epoch: 153, batch: 1, loss: 5.9103312492370605, modularity_loss: -0.5504608154296875, purity_loss: 6.381987571716309, regularization_loss: 0.07880452275276184 # 11:34:14 --- INFO: epoch: 154, batch: 1, loss: 5.881166458129883, modularity_loss: -0.5540558099746704, purity_loss: 6.356110572814941, regularization_loss: 0.07911158353090286 # 11:34:15 --- INFO: epoch: 155, batch: 1, loss: 5.85463285446167, modularity_loss: -0.5573135018348694, purity_loss: 6.33256196975708, regularization_loss: 0.0793842300772667 # 11:34:15 --- INFO: epoch: 156, batch: 1, loss: 5.831521987915039, modularity_loss: -0.560212254524231, purity_loss: 6.312124252319336, regularization_loss: 0.07961011677980423 # 11:34:15 --- INFO: epoch: 157, batch: 1, loss: 5.811868667602539, modularity_loss: -0.5627795457839966, purity_loss: 6.29486608505249, regularization_loss: 0.07978202402591705 # 11:34:15 --- INFO: epoch: 158, batch: 1, loss: 5.794981479644775, modularity_loss: -0.565090537071228, purity_loss: 6.280172824859619, regularization_loss: 0.07989932596683502 # 11:34:16 --- INFO: epoch: 159, batch: 1, loss: 5.779794216156006, modularity_loss: -0.5672242641448975, purity_loss: 6.267052173614502, regularization_loss: 0.07996627688407898 # 11:34:16 --- INFO: epoch: 160, batch: 1, loss: 5.7654242515563965, modularity_loss: -0.5692388415336609, purity_loss: 6.254673957824707, regularization_loss: 0.07998917996883392 # 11:34:16 --- INFO: epoch: 161, batch: 1, loss: 5.7513813972473145, modularity_loss: -0.5711544752120972, purity_loss: 6.242560386657715, regularization_loss: 0.07997548580169678 # 11:34:17 --- INFO: epoch: 162, batch: 1, loss: 5.737612247467041, modularity_loss: -0.572961688041687, purity_loss: 6.2306413650512695, regularization_loss: 0.07993237674236298 # 11:34:17 --- INFO: epoch: 163, batch: 1, loss: 5.724398612976074, modularity_loss: -0.5746380090713501, purity_loss: 6.219170093536377, regularization_loss: 0.07986671477556229 # 11:34:17 --- INFO: epoch: 164, batch: 1, loss: 5.712135314941406, modularity_loss: -0.5761581063270569, purity_loss: 6.2085089683532715, regularization_loss: 0.07978463917970657 # 11:34:18 --- INFO: epoch: 165, batch: 1, loss: 5.701193809509277, modularity_loss: -0.5775059461593628, purity_loss: 6.199007987976074, regularization_loss: 0.07969188690185547 # 11:34:18 --- INFO: epoch: 166, batch: 1, loss: 5.691784858703613, modularity_loss: -0.5786791443824768, purity_loss: 6.19087028503418, regularization_loss: 0.07959389686584473 # 11:34:18 --- INFO: epoch: 167, batch: 1, loss: 5.683870792388916, modularity_loss: -0.5796903371810913, purity_loss: 6.184064865112305, regularization_loss: 0.07949628680944443 # 11:34:18 --- INFO: epoch: 168, batch: 1, loss: 5.677199840545654, modularity_loss: -0.5805683732032776, purity_loss: 6.17836332321167, regularization_loss: 0.07940486073493958 # 11:34:19 --- INFO: epoch: 169, batch: 1, loss: 5.671406269073486, modularity_loss: -0.581353485584259, purity_loss: 6.173434734344482, regularization_loss: 0.07932505756616592 # 11:34:19 --- INFO: epoch: 170, batch: 1, loss: 5.666098594665527, modularity_loss: -0.5820875763893127, purity_loss: 6.168924808502197, regularization_loss: 0.07926131784915924 # 11:34:19 --- INFO: epoch: 171, batch: 1, loss: 5.66102933883667, modularity_loss: -0.58280348777771, purity_loss: 6.164616584777832, regularization_loss: 0.07921629399061203 # 11:34:20 --- INFO: epoch: 172, batch: 1, loss: 5.656072616577148, modularity_loss: -0.5835205316543579, purity_loss: 6.160402774810791, regularization_loss: 0.07919053733348846 # 11:34:20 --- INFO: epoch: 173, batch: 1, loss: 5.651222229003906, modularity_loss: -0.584238588809967, purity_loss: 6.156278610229492, regularization_loss: 0.07918201386928558 # 11:34:20 --- INFO: epoch: 174, batch: 1, loss: 5.64654541015625, modularity_loss: -0.5849401950836182, purity_loss: 6.152298450469971, regularization_loss: 0.07918698340654373 # 11:34:20 --- INFO: epoch: 175, batch: 1, loss: 5.642035007476807, modularity_loss: -0.585598349571228, purity_loss: 6.148432731628418, regularization_loss: 0.07920046895742416 # 11:34:21 --- INFO: epoch: 176, batch: 1, loss: 5.637692451477051, modularity_loss: -0.5861896276473999, purity_loss: 6.144664287567139, regularization_loss: 0.07921770215034485 # 11:34:21 --- INFO: epoch: 177, batch: 1, loss: 5.633500576019287, modularity_loss: -0.5866942405700684, purity_loss: 6.140960216522217, regularization_loss: 0.07923442870378494 # 11:34:21 --- INFO: epoch: 178, batch: 1, loss: 5.629417419433594, modularity_loss: -0.5871046781539917, purity_loss: 6.137274265289307, regularization_loss: 0.07924770563840866 # 11:34:22 --- INFO: epoch: 179, batch: 1, loss: 5.625455379486084, modularity_loss: -0.5874236822128296, purity_loss: 6.133623123168945, regularization_loss: 0.07925577461719513 # 11:34:22 --- INFO: epoch: 180, batch: 1, loss: 5.621650695800781, modularity_loss: -0.5876610279083252, purity_loss: 6.130053997039795, regularization_loss: 0.07925787568092346 # 11:34:22 --- INFO: epoch: 181, batch: 1, loss: 5.618027687072754, modularity_loss: -0.5878323316574097, purity_loss: 6.12660551071167, regularization_loss: 0.07925451546907425 # 11:34:23 --- INFO: epoch: 182, batch: 1, loss: 5.614570617675781, modularity_loss: -0.5879559516906738, purity_loss: 6.123280048370361, regularization_loss: 0.07924628257751465 # 11:34:23 --- INFO: epoch: 183, batch: 1, loss: 5.611252784729004, modularity_loss: -0.5880507826805115, purity_loss: 6.120069980621338, regularization_loss: 0.07923364639282227 # 11:34:23 --- INFO: epoch: 184, batch: 1, loss: 5.608041286468506, modularity_loss: -0.588132381439209, purity_loss: 6.1169562339782715, regularization_loss: 0.0792175903916359 # 11:34:24 --- INFO: epoch: 185, batch: 1, loss: 5.60488224029541, modularity_loss: -0.5882136225700378, purity_loss: 6.11389684677124, regularization_loss: 0.07919879257678986 # 11:34:24 --- INFO: epoch: 186, batch: 1, loss: 5.601746559143066, modularity_loss: -0.588301420211792, purity_loss: 6.110869884490967, regularization_loss: 0.07917796820402145 # 11:34:24 --- INFO: epoch: 187, batch: 1, loss: 5.598642349243164, modularity_loss: -0.5883959531784058, purity_loss: 6.107882976531982, regularization_loss: 0.07915551215410233 # 11:34:24 --- INFO: epoch: 188, batch: 1, loss: 5.5955986976623535, modularity_loss: -0.5884933471679688, purity_loss: 6.104959964752197, regularization_loss: 0.07913202047348022 # 11:34:25 --- INFO: epoch: 189, batch: 1, loss: 5.592626571655273, modularity_loss: -0.5885871052742004, purity_loss: 6.102105617523193, regularization_loss: 0.079107865691185 # 11:34:25 --- INFO: epoch: 190, batch: 1, loss: 5.589748382568359, modularity_loss: -0.5886681079864502, purity_loss: 6.0993332862854, regularization_loss: 0.07908322662115097 # 11:34:25 --- INFO: epoch: 191, batch: 1, loss: 5.5869646072387695, modularity_loss: -0.5887271165847778, purity_loss: 6.096633434295654, regularization_loss: 0.0790582224726677 # 11:34:26 --- INFO: epoch: 192, batch: 1, loss: 5.584257125854492, modularity_loss: -0.5887588262557983, purity_loss: 6.093983173370361, regularization_loss: 0.0790330022573471 # 11:34:26 --- INFO: epoch: 193, batch: 1, loss: 5.581605911254883, modularity_loss: -0.5887577533721924, purity_loss: 6.091355800628662, regularization_loss: 0.07900787144899368 # 11:34:26 --- INFO: epoch: 194, batch: 1, loss: 5.5789899826049805, modularity_loss: -0.5887237787246704, purity_loss: 6.088730812072754, regularization_loss: 0.07898299396038055 # 11:34:26 --- INFO: epoch: 195, batch: 1, loss: 5.576402187347412, modularity_loss: -0.5886590480804443, purity_loss: 6.086102485656738, regularization_loss: 0.07895872741937637 # 11:34:27 --- INFO: epoch: 196, batch: 1, loss: 5.573841571807861, modularity_loss: -0.5885686874389648, purity_loss: 6.083475112915039, regularization_loss: 0.07893519103527069 # 11:34:27 --- INFO: epoch: 197, batch: 1, loss: 5.571301460266113, modularity_loss: -0.5884584188461304, purity_loss: 6.08084774017334, regularization_loss: 0.07891237735748291 # 11:34:27 --- INFO: epoch: 198, batch: 1, loss: 5.568774223327637, modularity_loss: -0.588334321975708, purity_loss: 6.078218460083008, regularization_loss: 0.07889007031917572 # 11:34:28 --- INFO: epoch: 199, batch: 1, loss: 5.566244602203369, modularity_loss: -0.5882014036178589, purity_loss: 6.075578212738037, regularization_loss: 0.07886788249015808 # 11:34:28 --- INFO: epoch: 200, batch: 1, loss: 5.56368350982666, modularity_loss: -0.5880600214004517, purity_loss: 6.0728983879089355, regularization_loss: 0.0788450688123703 # 11:34:28 --- INFO: epoch: 201, batch: 1, loss: 5.561061382293701, modularity_loss: -0.587909460067749, purity_loss: 6.070149898529053, regularization_loss: 0.07882075756788254 # 11:34:29 --- INFO: epoch: 202, batch: 1, loss: 5.558346748352051, modularity_loss: -0.5877419710159302, purity_loss: 6.067295074462891, regularization_loss: 0.07879365235567093 # 11:34:29 --- INFO: epoch: 203, batch: 1, loss: 5.555512428283691, modularity_loss: -0.587546706199646, purity_loss: 6.064297199249268, regularization_loss: 0.07876210659742355 # 11:34:29 --- INFO: epoch: 204, batch: 1, loss: 5.5525221824646, modularity_loss: -0.5873087644577026, purity_loss: 6.0611066818237305, regularization_loss: 0.07872425019741058 # 11:34:29 --- INFO: epoch: 205, batch: 1, loss: 5.549304485321045, modularity_loss: -0.5870083570480347, purity_loss: 6.057635307312012, regularization_loss: 0.07867764681577682 # 11:34:30 --- INFO: epoch: 206, batch: 1, loss: 5.5457539558410645, modularity_loss: -0.5866184234619141, purity_loss: 6.05375337600708, regularization_loss: 0.07861891388893127 # 11:34:30 --- INFO: epoch: 207, batch: 1, loss: 5.541679382324219, modularity_loss: -0.5861009359359741, purity_loss: 6.04923677444458, regularization_loss: 0.07854343950748444 # 11:34:30 --- INFO: epoch: 208, batch: 1, loss: 5.536744594573975, modularity_loss: -0.5853960514068604, purity_loss: 6.043696880340576, regularization_loss: 0.07844388484954834 # 11:34:31 --- INFO: epoch: 209, batch: 1, loss: 5.530306339263916, modularity_loss: -0.5844054222106934, purity_loss: 6.036403656005859, regularization_loss: 0.07830830663442612 # 11:34:31 --- INFO: epoch: 210, batch: 1, loss: 5.52103328704834, modularity_loss: -0.5829585790634155, purity_loss: 6.025876998901367, regularization_loss: 0.07811501622200012 # 11:34:31 --- INFO: epoch: 211, batch: 1, loss: 5.505935192108154, modularity_loss: -0.5807530283927917, purity_loss: 6.008862495422363, regularization_loss: 0.07782554626464844 # 11:34:32 --- INFO: epoch: 212, batch: 1, loss: 5.4772233963012695, modularity_loss: -0.5772725343704224, purity_loss: 5.977132797241211, regularization_loss: 0.07736289501190186 # 11:34:32 --- INFO: epoch: 213, batch: 1, loss: 5.413817405700684, modularity_loss: -0.5718346238136292, purity_loss: 5.909071445465088, regularization_loss: 0.07658044248819351 # 11:34:32 --- INFO: epoch: 214, batch: 1, loss: 5.278060436248779, modularity_loss: -0.5637491941452026, purity_loss: 5.7665181159973145, regularization_loss: 0.07529158145189285 # 11:34:33 --- INFO: epoch: 215, batch: 1, loss: 5.112178802490234, modularity_loss: -0.5511707067489624, purity_loss: 5.589661121368408, regularization_loss: 0.0736885741353035 # 11:34:33 --- INFO: epoch: 216, batch: 1, loss: 5.020716190338135, modularity_loss: -0.5436804890632629, purity_loss: 5.491448879241943, regularization_loss: 0.07294774800539017 # 11:34:33 --- INFO: epoch: 217, batch: 1, loss: 4.824301242828369, modularity_loss: -0.5564278364181519, purity_loss: 5.307650089263916, regularization_loss: 0.07307903468608856 # 11:34:34 --- INFO: epoch: 218, batch: 1, loss: 4.668924808502197, modularity_loss: -0.5727019906044006, purity_loss: 5.168076992034912, regularization_loss: 0.07354971766471863 # 11:34:34 --- INFO: epoch: 219, batch: 1, loss: 4.594281196594238, modularity_loss: -0.5816538333892822, purity_loss: 5.102120876312256, regularization_loss: 0.07381407916545868 # 11:34:34 --- INFO: epoch: 220, batch: 1, loss: 4.531461238861084, modularity_loss: -0.5858439207077026, purity_loss: 5.0435357093811035, regularization_loss: 0.07376935333013535 # 11:34:35 --- INFO: epoch: 221, batch: 1, loss: 4.473836898803711, modularity_loss: -0.5879661440849304, purity_loss: 4.988247871398926, regularization_loss: 0.07355508208274841 # 11:34:35 --- INFO: epoch: 222, batch: 1, loss: 4.4354376792907715, modularity_loss: -0.5892443060874939, purity_loss: 4.951365947723389, regularization_loss: 0.07331585139036179 # 11:34:35 --- INFO: epoch: 223, batch: 1, loss: 4.4204607009887695, modularity_loss: -0.5902235507965088, purity_loss: 4.9375386238098145, regularization_loss: 0.07314546406269073 # 11:34:35 --- INFO: epoch: 224, batch: 1, loss: 4.420561790466309, modularity_loss: -0.5913387537002563, purity_loss: 4.938831329345703, regularization_loss: 0.07306916266679764 # 11:34:36 --- INFO: epoch: 225, batch: 1, loss: 4.423811435699463, modularity_loss: -0.5928112268447876, purity_loss: 4.943551063537598, regularization_loss: 0.07307147234678268 # 11:34:36 --- INFO: epoch: 226, batch: 1, loss: 4.422661304473877, modularity_loss: -0.5944815874099731, purity_loss: 4.944013595581055, regularization_loss: 0.07312919944524765 # 11:34:36 --- INFO: epoch: 227, batch: 1, loss: 4.4168314933776855, modularity_loss: -0.596046507358551, purity_loss: 4.939652442932129, regularization_loss: 0.0732254907488823 # 11:34:37 --- INFO: epoch: 228, batch: 1, loss: 4.410677433013916, modularity_loss: -0.5972778797149658, purity_loss: 4.934606075286865, regularization_loss: 0.07334909588098526 # 11:34:37 --- INFO: epoch: 229, batch: 1, loss: 4.408371448516846, modularity_loss: -0.5980930328369141, purity_loss: 4.932977676391602, regularization_loss: 0.07348683476448059 # 11:34:37 --- INFO: epoch: 230, batch: 1, loss: 4.409975528717041, modularity_loss: -0.5985338687896729, purity_loss: 4.934889793395996, regularization_loss: 0.07361973822116852 # 11:34:38 --- INFO: epoch: 231, batch: 1, loss: 4.411879539489746, modularity_loss: -0.5987254977226257, purity_loss: 4.936878204345703, regularization_loss: 0.07372687011957169 # 11:34:38 --- INFO: epoch: 232, batch: 1, loss: 4.411393165588379, modularity_loss: -0.598797082901001, purity_loss: 4.936393737792969, regularization_loss: 0.0737965852022171 # 11:34:38 --- INFO: epoch: 233, batch: 1, loss: 4.409461975097656, modularity_loss: -0.598831295967102, purity_loss: 4.93446159362793, regularization_loss: 0.07383144646883011 # 11:34:39 --- INFO: epoch: 234, batch: 1, loss: 4.408030986785889, modularity_loss: -0.5988819599151611, purity_loss: 4.933070659637451, regularization_loss: 0.07384243607521057 # 11:34:39 --- INFO: epoch: 235, batch: 1, loss: 4.40681266784668, modularity_loss: -0.5990049242973328, purity_loss: 4.931978225708008, regularization_loss: 0.07383940368890762 # 11:34:39 --- INFO: epoch: 236, batch: 1, loss: 4.404171466827393, modularity_loss: -0.5992487668991089, purity_loss: 4.929592609405518, regularization_loss: 0.07382772862911224 # 11:34:40 --- INFO: epoch: 237, batch: 1, loss: 4.39953088760376, modularity_loss: -0.5996174812316895, purity_loss: 4.9253387451171875, regularization_loss: 0.07380976527929306 # 11:34:40 --- INFO: epoch: 238, batch: 1, loss: 4.394030570983887, modularity_loss: -0.6000624895095825, purity_loss: 4.92030668258667, regularization_loss: 0.0737866461277008 # 11:34:40 --- INFO: epoch: 239, batch: 1, loss: 4.389148712158203, modularity_loss: -0.6005113124847412, purity_loss: 4.915902614593506, regularization_loss: 0.07375739514827728 # 11:34:41 --- INFO: epoch: 240, batch: 1, loss: 4.385226249694824, modularity_loss: -0.6008815765380859, purity_loss: 4.912387371063232, regularization_loss: 0.07372060418128967 # 11:34:41 --- INFO: epoch: 241, batch: 1, loss: 4.381584167480469, modularity_loss: -0.60111004114151, purity_loss: 4.9090189933776855, regularization_loss: 0.07367543876171112 # 11:34:41 --- INFO: epoch: 242, batch: 1, loss: 4.377500057220459, modularity_loss: -0.601157546043396, purity_loss: 4.905034065246582, regularization_loss: 0.07362337410449982 # 11:34:42 --- INFO: epoch: 243, batch: 1, loss: 4.3730010986328125, modularity_loss: -0.6010140180587769, purity_loss: 4.900446891784668, regularization_loss: 0.07356845587491989 # 11:34:42 --- INFO: epoch: 244, batch: 1, loss: 4.368607521057129, modularity_loss: -0.6007052659988403, purity_loss: 4.895796298980713, regularization_loss: 0.07351662963628769 # 11:34:42 --- INFO: epoch: 245, batch: 1, loss: 4.364609718322754, modularity_loss: -0.6002802848815918, purity_loss: 4.891416549682617, regularization_loss: 0.07347341626882553 # 11:34:42 --- INFO: epoch: 246, batch: 1, loss: 4.360860824584961, modularity_loss: -0.5997978448867798, purity_loss: 4.887216567993164, regularization_loss: 0.07344198226928711 # 11:34:43 --- INFO: epoch: 247, batch: 1, loss: 4.356934070587158, modularity_loss: -0.5993043184280396, purity_loss: 4.882814884185791, regularization_loss: 0.07342366874217987 # 11:34:43 --- INFO: epoch: 248, batch: 1, loss: 4.35261869430542, modularity_loss: -0.5988279581069946, purity_loss: 4.87802791595459, regularization_loss: 0.07341862469911575 # 11:34:43 --- INFO: epoch: 249, batch: 1, loss: 4.348060131072998, modularity_loss: -0.5983699560165405, purity_loss: 4.87300443649292, regularization_loss: 0.07342550158500671 # 11:34:44 --- INFO: epoch: 250, batch: 1, loss: 4.343605041503906, modularity_loss: -0.5979124307632446, purity_loss: 4.868075847625732, regularization_loss: 0.07344188541173935 # 11:34:44 --- INFO: epoch: 251, batch: 1, loss: 4.339545726776123, modularity_loss: -0.5974299907684326, purity_loss: 4.863511562347412, regularization_loss: 0.07346434146165848 # 11:34:44 --- INFO: epoch: 252, batch: 1, loss: 4.33583402633667, modularity_loss: -0.5968954563140869, purity_loss: 4.859240531921387, regularization_loss: 0.07348891347646713 # 11:34:45 --- INFO: epoch: 253, batch: 1, loss: 4.332301139831543, modularity_loss: -0.5962988138198853, purity_loss: 4.855088233947754, regularization_loss: 0.07351180166006088 # 11:34:45 --- INFO: epoch: 254, batch: 1, loss: 4.328784942626953, modularity_loss: -0.5956467390060425, purity_loss: 4.8509016036987305, regularization_loss: 0.0735301673412323 # 11:34:45 --- INFO: epoch: 255, batch: 1, loss: 4.325329780578613, modularity_loss: -0.5949624180793762, purity_loss: 4.846749782562256, regularization_loss: 0.07354238629341125 # 11:34:45 --- INFO: epoch: 256, batch: 1, loss: 4.322075843811035, modularity_loss: -0.5942855477333069, purity_loss: 4.842813491821289, regularization_loss: 0.07354779541492462 # 11:34:46 --- INFO: epoch: 257, batch: 1, loss: 4.319091320037842, modularity_loss: -0.5936647653579712, purity_loss: 4.839209079742432, regularization_loss: 0.0735468715429306 # 11:34:46 --- INFO: epoch: 258, batch: 1, loss: 4.316356658935547, modularity_loss: -0.5931333303451538, purity_loss: 4.835948944091797, regularization_loss: 0.07354124635457993 # 11:34:46 --- INFO: epoch: 259, batch: 1, loss: 4.313860893249512, modularity_loss: -0.5927048921585083, purity_loss: 4.833033084869385, regularization_loss: 0.07353287190198898 # 11:34:47 --- INFO: epoch: 260, batch: 1, loss: 4.311649799346924, modularity_loss: -0.5923765897750854, purity_loss: 4.830502510070801, regularization_loss: 0.07352384179830551 # 11:34:47 --- INFO: epoch: 261, batch: 1, loss: 4.309781074523926, modularity_loss: -0.5921224355697632, purity_loss: 4.828387260437012, regularization_loss: 0.07351619750261307 # 11:34:47 --- INFO: epoch: 262, batch: 1, loss: 4.30827522277832, modularity_loss: -0.5919064283370972, purity_loss: 4.8266706466674805, regularization_loss: 0.07351125776767731 # 11:34:48 --- INFO: epoch: 263, batch: 1, loss: 4.307024955749512, modularity_loss: -0.5916910171508789, purity_loss: 4.825206279754639, regularization_loss: 0.07350989431142807 # 11:34:48 --- INFO: epoch: 264, batch: 1, loss: 4.305865287780762, modularity_loss: -0.5914521217346191, purity_loss: 4.82380485534668, regularization_loss: 0.07351233810186386 # 11:34:48 --- INFO: epoch: 265, batch: 1, loss: 4.304628849029541, modularity_loss: -0.5911808013916016, purity_loss: 4.822291374206543, regularization_loss: 0.07351834326982498 # 11:34:49 --- INFO: epoch: 266, batch: 1, loss: 4.303272724151611, modularity_loss: -0.5908845663070679, purity_loss: 4.820630073547363, regularization_loss: 0.07352729886770248 # 11:34:49 --- INFO: epoch: 267, batch: 1, loss: 4.301836967468262, modularity_loss: -0.5905880331993103, purity_loss: 4.818886756896973, regularization_loss: 0.07353837788105011 # 11:34:49 --- INFO: epoch: 268, batch: 1, loss: 4.3003973960876465, modularity_loss: -0.5903263688087463, purity_loss: 4.817173480987549, regularization_loss: 0.07355019450187683 # 11:34:49 --- INFO: epoch: 269, batch: 1, loss: 4.298980712890625, modularity_loss: -0.5901319980621338, purity_loss: 4.8155517578125, regularization_loss: 0.07356119155883789 # 11:34:50 --- INFO: epoch: 270, batch: 1, loss: 4.297562122344971, modularity_loss: -0.5900236368179321, purity_loss: 4.814016342163086, regularization_loss: 0.07356960326433182 # 11:34:50 --- INFO: epoch: 271, batch: 1, loss: 4.2961015701293945, modularity_loss: -0.5900083780288696, purity_loss: 4.812536239624023, regularization_loss: 0.07357395440340042 # 11:34:50 --- INFO: epoch: 272, batch: 1, loss: 4.294596195220947, modularity_loss: -0.5900752544403076, purity_loss: 4.811098098754883, regularization_loss: 0.07357331365346909 # 11:34:51 --- INFO: epoch: 273, batch: 1, loss: 4.293078422546387, modularity_loss: -0.5902026891708374, purity_loss: 4.809713840484619, regularization_loss: 0.07356734573841095 # 11:34:51 --- INFO: epoch: 274, batch: 1, loss: 4.291578769683838, modularity_loss: -0.5903612375259399, purity_loss: 4.808383464813232, regularization_loss: 0.07355649024248123 # 11:34:51 --- INFO: epoch: 275, batch: 1, loss: 4.290128231048584, modularity_loss: -0.5905221700668335, purity_loss: 4.8071088790893555, regularization_loss: 0.07354172319173813 # 11:34:52 --- INFO: epoch: 276, batch: 1, loss: 4.2887420654296875, modularity_loss: -0.5906651616096497, purity_loss: 4.805882930755615, regularization_loss: 0.07352451980113983 # 11:34:52 --- INFO: epoch: 277, batch: 1, loss: 4.287437438964844, modularity_loss: -0.5907781720161438, purity_loss: 4.804708957672119, regularization_loss: 0.07350655645132065 # 11:34:52 --- INFO: epoch: 278, batch: 1, loss: 4.286211967468262, modularity_loss: -0.5908651351928711, purity_loss: 4.803586959838867, regularization_loss: 0.07348987460136414 # 11:34:52 --- INFO: epoch: 279, batch: 1, loss: 4.28505802154541, modularity_loss: -0.5909320116043091, purity_loss: 4.80251407623291, regularization_loss: 0.07347596436738968 # 11:34:53 --- INFO: epoch: 280, batch: 1, loss: 4.283960342407227, modularity_loss: -0.5909901857376099, purity_loss: 4.80148458480835, regularization_loss: 0.07346581667661667 # 11:34:53 --- INFO: epoch: 281, batch: 1, loss: 4.282896518707275, modularity_loss: -0.5910468101501465, purity_loss: 4.800483703613281, regularization_loss: 0.0734596699476242 # 11:34:53 --- INFO: epoch: 282, batch: 1, loss: 4.281870365142822, modularity_loss: -0.5911058783531189, purity_loss: 4.799519062042236, regularization_loss: 0.07345734536647797 # 11:34:54 --- INFO: epoch: 283, batch: 1, loss: 4.280886650085449, modularity_loss: -0.5911659002304077, purity_loss: 4.7985944747924805, regularization_loss: 0.07345793396234512 # 11:34:54 --- INFO: epoch: 284, batch: 1, loss: 4.279959678649902, modularity_loss: -0.591220498085022, purity_loss: 4.797719955444336, regularization_loss: 0.07346030324697495 # 11:34:54 --- INFO: epoch: 285, batch: 1, loss: 4.279078006744385, modularity_loss: -0.5912607908248901, purity_loss: 4.796875476837158, regularization_loss: 0.07346325367689133 # 11:34:55 --- INFO: epoch: 286, batch: 1, loss: 4.278225421905518, modularity_loss: -0.5912783741950989, purity_loss: 4.7960381507873535, regularization_loss: 0.07346563786268234 # 11:34:55 --- INFO: epoch: 287, batch: 1, loss: 4.277387619018555, modularity_loss: -0.5912687182426453, purity_loss: 4.795189380645752, regularization_loss: 0.07346677035093307 # 11:34:55 --- INFO: epoch: 288, batch: 1, loss: 4.276552677154541, modularity_loss: -0.5912315845489502, purity_loss: 4.794317722320557, regularization_loss: 0.073466457426548 # 11:34:56 --- INFO: epoch: 289, batch: 1, loss: 4.2757248878479, modularity_loss: -0.5911734104156494, purity_loss: 4.79343318939209, regularization_loss: 0.0734650120139122 # 11:34:56 --- INFO: epoch: 290, batch: 1, loss: 4.274920463562012, modularity_loss: -0.5911019444465637, purity_loss: 4.7925591468811035, regularization_loss: 0.07346304506063461 # 11:34:56 --- INFO: epoch: 291, batch: 1, loss: 4.274134635925293, modularity_loss: -0.5910259485244751, purity_loss: 4.791699409484863, regularization_loss: 0.07346127927303314 # 11:34:57 --- INFO: epoch: 292, batch: 1, loss: 4.2733635902404785, modularity_loss: -0.5909523963928223, purity_loss: 4.790855407714844, regularization_loss: 0.07346054166555405 # 11:34:57 --- INFO: epoch: 293, batch: 1, loss: 4.272598743438721, modularity_loss: -0.5908820629119873, purity_loss: 4.790019512176514, regularization_loss: 0.07346116006374359 # 11:34:57 --- INFO: epoch: 294, batch: 1, loss: 4.271836757659912, modularity_loss: -0.5908104181289673, purity_loss: 4.789184093475342, regularization_loss: 0.07346310466527939 # 11:34:57 --- INFO: epoch: 295, batch: 1, loss: 4.2710700035095215, modularity_loss: -0.5907310247421265, purity_loss: 4.788334846496582, regularization_loss: 0.07346601039171219 # 11:34:58 --- INFO: epoch: 296, batch: 1, loss: 4.270299434661865, modularity_loss: -0.5906384587287903, purity_loss: 4.787468433380127, regularization_loss: 0.07346942275762558 # 11:34:58 --- INFO: epoch: 297, batch: 1, loss: 4.269519805908203, modularity_loss: -0.5905280113220215, purity_loss: 4.7865753173828125, regularization_loss: 0.07347261905670166 # 11:34:58 --- INFO: epoch: 298, batch: 1, loss: 4.268729209899902, modularity_loss: -0.5903987288475037, purity_loss: 4.785653114318848, regularization_loss: 0.07347502559423447 # 11:34:59 --- INFO: epoch: 299, batch: 1, loss: 4.267923831939697, modularity_loss: -0.590255618095398, purity_loss: 4.784703254699707, regularization_loss: 0.07347619533538818 # 11:34:59 --- INFO: epoch: 300, batch: 1, loss: 4.2671027183532715, modularity_loss: -0.5901046991348267, purity_loss: 4.783731460571289, regularization_loss: 0.0734758973121643 # 11:34:59 --- INFO: epoch: 301, batch: 1, loss: 4.266260147094727, modularity_loss: -0.589952826499939, purity_loss: 4.78273868560791, regularization_loss: 0.07347415387630463 # 11:35:00 --- INFO: epoch: 302, batch: 1, loss: 4.265389442443848, modularity_loss: -0.5898054838180542, purity_loss: 4.781723976135254, regularization_loss: 0.07347114384174347 # 11:35:00 --- INFO: epoch: 303, batch: 1, loss: 4.264482021331787, modularity_loss: -0.5896624326705933, purity_loss: 4.780677318572998, regularization_loss: 0.07346729934215546 # 11:35:00 --- INFO: epoch: 304, batch: 1, loss: 4.263533115386963, modularity_loss: -0.589518666267395, purity_loss: 4.77958869934082, regularization_loss: 0.07346303761005402 # 11:35:01 --- INFO: epoch: 305, batch: 1, loss: 4.262538909912109, modularity_loss: -0.5893668532371521, purity_loss: 4.778447151184082, regularization_loss: 0.07345866411924362 # 11:35:01 --- INFO: epoch: 306, batch: 1, loss: 4.261477470397949, modularity_loss: -0.589195966720581, purity_loss: 4.777218818664551, regularization_loss: 0.07345455139875412 # 11:35:02 --- INFO: epoch: 307, batch: 1, loss: 4.260337829589844, modularity_loss: -0.5889950394630432, purity_loss: 4.775881767272949, regularization_loss: 0.07345087826251984 # 11:35:02 --- INFO: epoch: 308, batch: 1, loss: 4.259085655212402, modularity_loss: -0.5887542963027954, purity_loss: 4.774392127990723, regularization_loss: 0.0734478235244751 # 11:35:02 --- INFO: epoch: 309, batch: 1, loss: 4.257706165313721, modularity_loss: -0.5884677767753601, purity_loss: 4.772728443145752, regularization_loss: 0.07344553619623184 # 11:35:02 --- INFO: epoch: 310, batch: 1, loss: 4.256180286407471, modularity_loss: -0.5881308913230896, purity_loss: 4.770867347717285, regularization_loss: 0.0734439268708229 # 11:35:03 --- INFO: epoch: 311, batch: 1, loss: 4.254464149475098, modularity_loss: -0.5877430438995361, purity_loss: 4.768764019012451, regularization_loss: 0.07344336062669754 # 11:35:03 --- INFO: epoch: 312, batch: 1, loss: 4.252506256103516, modularity_loss: -0.5872955918312073, purity_loss: 4.766357898712158, regularization_loss: 0.07344380021095276 # 11:35:04 --- INFO: epoch: 313, batch: 1, loss: 4.250242710113525, modularity_loss: -0.5867732167243958, purity_loss: 4.763570785522461, regularization_loss: 0.07344522327184677 # 11:35:04 --- INFO: epoch: 314, batch: 1, loss: 4.247618675231934, modularity_loss: -0.5861533284187317, purity_loss: 4.760324001312256, regularization_loss: 0.07344774156808853 # 11:35:04 --- INFO: epoch: 315, batch: 1, loss: 4.2445502281188965, modularity_loss: -0.5854088068008423, purity_loss: 4.75650691986084, regularization_loss: 0.07345198094844818 # 11:35:04 --- INFO: epoch: 316, batch: 1, loss: 4.240986347198486, modularity_loss: -0.5845093727111816, purity_loss: 4.752037048339844, regularization_loss: 0.07345881313085556 # 11:35:05 --- INFO: epoch: 317, batch: 1, loss: 4.236891269683838, modularity_loss: -0.5834317207336426, purity_loss: 4.746853351593018, regularization_loss: 0.07346955686807632 # 11:35:05 --- INFO: epoch: 318, batch: 1, loss: 4.232332706451416, modularity_loss: -0.5821704864501953, purity_loss: 4.7410173416137695, regularization_loss: 0.07348573207855225 # 11:35:05 --- INFO: epoch: 319, batch: 1, loss: 4.227492332458496, modularity_loss: -0.5807698965072632, purity_loss: 4.734753131866455, regularization_loss: 0.07350895553827286 # 11:35:06 --- INFO: epoch: 320, batch: 1, loss: 4.222650051116943, modularity_loss: -0.5793299674987793, purity_loss: 4.728439807891846, regularization_loss: 0.07354001700878143 # 11:35:06 --- INFO: epoch: 321, batch: 1, loss: 4.218080997467041, modularity_loss: -0.5780047178268433, purity_loss: 4.722506999969482, regularization_loss: 0.07357867062091827 # 11:35:06 --- INFO: epoch: 322, batch: 1, loss: 4.213959217071533, modularity_loss: -0.5769439935684204, purity_loss: 4.71727991104126, regularization_loss: 0.0736234113574028 # 11:35:07 --- INFO: epoch: 323, batch: 1, loss: 4.210313320159912, modularity_loss: -0.5762346982955933, purity_loss: 4.712876796722412, regularization_loss: 0.07367132604122162 # 11:35:07 --- INFO: epoch: 324, batch: 1, loss: 4.207118511199951, modularity_loss: -0.5758606791496277, purity_loss: 4.709259033203125, regularization_loss: 0.07372003048658371 # 11:35:07 --- INFO: epoch: 325, batch: 1, loss: 4.204360485076904, modularity_loss: -0.575730562210083, purity_loss: 4.70632266998291, regularization_loss: 0.07376838475465775 # 11:35:08 --- INFO: epoch: 326, batch: 1, loss: 4.201966762542725, modularity_loss: -0.5757330656051636, purity_loss: 4.703883647918701, regularization_loss: 0.07381632924079895 # 11:35:08 --- INFO: epoch: 327, batch: 1, loss: 4.199801445007324, modularity_loss: -0.5757776498794556, purity_loss: 4.701714515686035, regularization_loss: 0.07386460900306702 # 11:35:08 --- INFO: epoch: 328, batch: 1, loss: 4.197755813598633, modularity_loss: -0.5758078098297119, purity_loss: 4.699650287628174, regularization_loss: 0.07391347736120224 # 11:35:09 --- INFO: epoch: 329, batch: 1, loss: 4.195771217346191, modularity_loss: -0.5758171081542969, purity_loss: 4.697626113891602, regularization_loss: 0.07396231591701508 # 11:35:09 --- INFO: epoch: 330, batch: 1, loss: 4.193813323974609, modularity_loss: -0.5758206248283386, purity_loss: 4.695624351501465, regularization_loss: 0.07400976866483688 # 11:35:09 --- INFO: epoch: 331, batch: 1, loss: 4.191867351531982, modularity_loss: -0.5758447647094727, purity_loss: 4.693657875061035, regularization_loss: 0.07405406981706619 # 11:35:10 --- INFO: epoch: 332, batch: 1, loss: 4.189898490905762, modularity_loss: -0.5759114027023315, purity_loss: 4.691716194152832, regularization_loss: 0.07409362494945526 # 11:35:10 --- INFO: epoch: 333, batch: 1, loss: 4.187904357910156, modularity_loss: -0.5760267972946167, purity_loss: 4.689803600311279, regularization_loss: 0.07412758469581604 # 11:35:10 --- INFO: epoch: 334, batch: 1, loss: 4.185907363891602, modularity_loss: -0.5761787295341492, purity_loss: 4.687930107116699, regularization_loss: 0.07415594160556793 # 11:35:11 --- INFO: epoch: 335, batch: 1, loss: 4.183969974517822, modularity_loss: -0.5763427019119263, purity_loss: 4.68613338470459, regularization_loss: 0.07417944818735123 # 11:35:11 --- INFO: epoch: 336, batch: 1, loss: 4.182136535644531, modularity_loss: -0.5764858722686768, purity_loss: 4.684423446655273, regularization_loss: 0.07419916987419128 # 11:35:11 --- INFO: epoch: 337, batch: 1, loss: 4.180404186248779, modularity_loss: -0.576573371887207, purity_loss: 4.682761192321777, regularization_loss: 0.07421652227640152 # 11:35:12 --- INFO: epoch: 338, batch: 1, loss: 4.178768157958984, modularity_loss: -0.5765821933746338, purity_loss: 4.681118011474609, regularization_loss: 0.07423239946365356 # 11:35:12 --- INFO: epoch: 339, batch: 1, loss: 4.177211284637451, modularity_loss: -0.576517641544342, purity_loss: 4.679482460021973, regularization_loss: 0.07424658536911011 # 11:35:12 --- INFO: epoch: 340, batch: 1, loss: 4.175746440887451, modularity_loss: -0.5763891339302063, purity_loss: 4.677876949310303, regularization_loss: 0.0742587149143219 # 11:35:12 --- INFO: epoch: 341, batch: 1, loss: 4.174374580383301, modularity_loss: -0.5762315392494202, purity_loss: 4.6763386726379395, regularization_loss: 0.07426769286394119 # 11:35:13 --- INFO: epoch: 342, batch: 1, loss: 4.17310094833374, modularity_loss: -0.5760815739631653, purity_loss: 4.674910068511963, regularization_loss: 0.07427257299423218 # 11:35:13 --- INFO: epoch: 343, batch: 1, loss: 4.171891689300537, modularity_loss: -0.5759718418121338, purity_loss: 4.673590660095215, regularization_loss: 0.07427293807268143 # 11:35:13 --- INFO: epoch: 344, batch: 1, loss: 4.170742511749268, modularity_loss: -0.5759111642837524, purity_loss: 4.672384262084961, regularization_loss: 0.07426943629980087 # 11:35:14 --- INFO: epoch: 345, batch: 1, loss: 4.169649600982666, modularity_loss: -0.5758869647979736, purity_loss: 4.6712727546691895, regularization_loss: 0.07426375895738602 # 11:35:14 --- INFO: epoch: 346, batch: 1, loss: 4.168609142303467, modularity_loss: -0.5758694410324097, purity_loss: 4.670220375061035, regularization_loss: 0.0742582380771637 # 11:35:14 --- INFO: epoch: 347, batch: 1, loss: 4.167614936828613, modularity_loss: -0.5758272409439087, purity_loss: 4.669187068939209, regularization_loss: 0.0742550790309906 # 11:35:15 --- INFO: epoch: 348, batch: 1, loss: 4.166640281677246, modularity_loss: -0.5757341980934143, purity_loss: 4.668118476867676, regularization_loss: 0.07425595074892044 # 11:35:15 --- INFO: epoch: 349, batch: 1, loss: 4.165668964385986, modularity_loss: -0.5755927562713623, purity_loss: 4.6670002937316895, regularization_loss: 0.07426134496927261 # 11:35:15 --- INFO: epoch: 350, batch: 1, loss: 4.164701461791992, modularity_loss: -0.5754252672195435, purity_loss: 4.665855884552002, regularization_loss: 0.07427066564559937 # 11:35:16 --- INFO: epoch: 351, batch: 1, loss: 4.163743019104004, modularity_loss: -0.5752644538879395, purity_loss: 4.664724826812744, regularization_loss: 0.07428259402513504 # 11:35:16 --- INFO: epoch: 352, batch: 1, loss: 4.162790775299072, modularity_loss: -0.5751410722732544, purity_loss: 4.663636207580566, regularization_loss: 0.07429562509059906 # 11:35:16 --- INFO: epoch: 353, batch: 1, loss: 4.161850929260254, modularity_loss: -0.5750676393508911, purity_loss: 4.662610054016113, regularization_loss: 0.07430870085954666 # 11:35:17 --- INFO: epoch: 354, batch: 1, loss: 4.160918712615967, modularity_loss: -0.5750381946563721, purity_loss: 4.661635398864746, regularization_loss: 0.074321448802948 # 11:35:17 --- INFO: epoch: 355, batch: 1, loss: 4.160010814666748, modularity_loss: -0.5750309228897095, purity_loss: 4.660707473754883, regularization_loss: 0.07433409243822098 # 11:35:17 --- INFO: epoch: 356, batch: 1, loss: 4.159125328063965, modularity_loss: -0.5750200748443604, purity_loss: 4.6597981452941895, regularization_loss: 0.07434719800949097 # 11:35:18 --- INFO: epoch: 357, batch: 1, loss: 4.158264636993408, modularity_loss: -0.5749863982200623, purity_loss: 4.6588897705078125, regularization_loss: 0.07436109334230423 # 11:35:18 --- INFO: epoch: 358, batch: 1, loss: 4.157419204711914, modularity_loss: -0.5749261379241943, purity_loss: 4.6579694747924805, regularization_loss: 0.07437564432621002 # 11:35:18 --- INFO: epoch: 359, batch: 1, loss: 4.1566033363342285, modularity_loss: -0.5748509168624878, purity_loss: 4.657063961029053, regularization_loss: 0.0743902176618576 # 11:35:18 --- INFO: epoch: 360, batch: 1, loss: 4.155808448791504, modularity_loss: -0.5747811794281006, purity_loss: 4.656185626983643, regularization_loss: 0.07440382242202759 # 11:35:19 --- INFO: epoch: 361, batch: 1, loss: 4.1550397872924805, modularity_loss: -0.5747342109680176, purity_loss: 4.65535831451416, regularization_loss: 0.0744156464934349 # 11:35:19 --- INFO: epoch: 362, batch: 1, loss: 4.154287815093994, modularity_loss: -0.57471764087677, purity_loss: 4.654580116271973, regularization_loss: 0.0744253620505333 # 11:35:19 --- INFO: epoch: 363, batch: 1, loss: 4.1535539627075195, modularity_loss: -0.574722945690155, purity_loss: 4.653843402862549, regularization_loss: 0.0744333267211914 # 11:35:20 --- INFO: epoch: 364, batch: 1, loss: 4.152837753295898, modularity_loss: -0.5747337937355042, purity_loss: 4.653131484985352, regularization_loss: 0.07444030791521072 # 11:35:20 --- INFO: epoch: 365, batch: 1, loss: 4.152139663696289, modularity_loss: -0.5747319459915161, purity_loss: 4.6524248123168945, regularization_loss: 0.07444706559181213 # 11:35:20 --- INFO: epoch: 366, batch: 1, loss: 4.151451110839844, modularity_loss: -0.5747050046920776, purity_loss: 4.651701927185059, regularization_loss: 0.07445415109395981 # 11:35:21 --- INFO: epoch: 367, batch: 1, loss: 4.1507697105407715, modularity_loss: -0.5746498107910156, purity_loss: 4.650958061218262, regularization_loss: 0.07446164637804031 # 11:35:21 --- INFO: epoch: 368, batch: 1, loss: 4.1500959396362305, modularity_loss: -0.5745770931243896, purity_loss: 4.650203704833984, regularization_loss: 0.07446911931037903 # 11:35:21 --- INFO: epoch: 369, batch: 1, loss: 4.149426460266113, modularity_loss: -0.574503481388092, purity_loss: 4.649454116821289, regularization_loss: 0.07447592914104462 # 11:35:21 --- INFO: epoch: 370, batch: 1, loss: 4.148763179779053, modularity_loss: -0.5744418501853943, purity_loss: 4.648723602294922, regularization_loss: 0.07448150962591171 # 11:35:22 --- INFO: epoch: 371, batch: 1, loss: 4.148103713989258, modularity_loss: -0.5743969678878784, purity_loss: 4.648015022277832, regularization_loss: 0.07448570430278778 # 11:35:22 --- INFO: epoch: 372, batch: 1, loss: 4.14744758605957, modularity_loss: -0.5743646025657654, purity_loss: 4.647323131561279, regularization_loss: 0.07448887079954147 # 11:35:22 --- INFO: epoch: 373, batch: 1, loss: 4.1468000411987305, modularity_loss: -0.5743318796157837, purity_loss: 4.646640300750732, regularization_loss: 0.07449159026145935 # 11:35:23 --- INFO: epoch: 374, batch: 1, loss: 4.146157264709473, modularity_loss: -0.5742868185043335, purity_loss: 4.645949363708496, regularization_loss: 0.07449451833963394 # 11:35:23 --- INFO: epoch: 375, batch: 1, loss: 4.145514011383057, modularity_loss: -0.5742236971855164, purity_loss: 4.64523983001709, regularization_loss: 0.07449795305728912 # 11:35:23 --- INFO: epoch: 376, batch: 1, loss: 4.144867897033691, modularity_loss: -0.5741462111473083, purity_loss: 4.644512176513672, regularization_loss: 0.0745018869638443 # 11:35:24 --- INFO: epoch: 377, batch: 1, loss: 4.144218921661377, modularity_loss: -0.5740619897842407, purity_loss: 4.64377498626709, regularization_loss: 0.07450602948665619 # 11:35:24 --- INFO: epoch: 378, batch: 1, loss: 4.1435675621032715, modularity_loss: -0.5739832520484924, purity_loss: 4.643040657043457, regularization_loss: 0.07450997829437256 # 11:35:24 --- INFO: epoch: 379, batch: 1, loss: 4.14290714263916, modularity_loss: -0.5739157795906067, purity_loss: 4.642309188842773, regularization_loss: 0.07451354712247849 # 11:35:25 --- INFO: epoch: 380, batch: 1, loss: 4.142237663269043, modularity_loss: -0.5738581418991089, purity_loss: 4.641578674316406, regularization_loss: 0.07451687753200531 # 11:35:25 --- INFO: epoch: 381, batch: 1, loss: 4.141550064086914, modularity_loss: -0.5738012194633484, purity_loss: 4.6408305168151855, regularization_loss: 0.0745205283164978 # 11:35:25 --- INFO: epoch: 382, batch: 1, loss: 4.140843868255615, modularity_loss: -0.5737334489822388, purity_loss: 4.640052318572998, regularization_loss: 0.07452511042356491 # 11:35:26 --- INFO: epoch: 383, batch: 1, loss: 4.140113353729248, modularity_loss: -0.5736440420150757, purity_loss: 4.63922643661499, regularization_loss: 0.07453113049268723 # 11:35:26 --- INFO: epoch: 384, batch: 1, loss: 4.139350414276123, modularity_loss: -0.5735290050506592, purity_loss: 4.638340473175049, regularization_loss: 0.07453881949186325 # 11:35:26 --- INFO: epoch: 385, batch: 1, loss: 4.13855504989624, modularity_loss: -0.5733901858329773, purity_loss: 4.637397289276123, regularization_loss: 0.07454805821180344 # 11:35:27 --- INFO: epoch: 386, batch: 1, loss: 4.137716293334961, modularity_loss: -0.5732322931289673, purity_loss: 4.63638973236084, regularization_loss: 0.07455860078334808 # 11:35:27 --- INFO: epoch: 387, batch: 1, loss: 4.136832237243652, modularity_loss: -0.5730576515197754, purity_loss: 4.635319709777832, regularization_loss: 0.0745701938867569 # 11:35:27 --- INFO: epoch: 388, batch: 1, loss: 4.135908603668213, modularity_loss: -0.5728645920753479, purity_loss: 4.634190559387207, regularization_loss: 0.07458268105983734 # 11:35:27 --- INFO: epoch: 389, batch: 1, loss: 4.134945869445801, modularity_loss: -0.5726447105407715, purity_loss: 4.632994174957275, regularization_loss: 0.07459626346826553 # 11:35:28 --- INFO: epoch: 390, batch: 1, loss: 4.133967876434326, modularity_loss: -0.572390079498291, purity_loss: 4.631746768951416, regularization_loss: 0.07461108267307281 # 11:35:28 --- INFO: epoch: 391, batch: 1, loss: 4.133007526397705, modularity_loss: -0.5721008777618408, purity_loss: 4.630481243133545, regularization_loss: 0.07462704181671143 # 11:35:28 --- INFO: epoch: 392, batch: 1, loss: 4.132107734680176, modularity_loss: -0.5717844367027283, purity_loss: 4.62924861907959, regularization_loss: 0.07464379072189331 # 11:35:29 --- INFO: epoch: 393, batch: 1, loss: 4.131323337554932, modularity_loss: -0.5714603066444397, purity_loss: 4.6281232833862305, regularization_loss: 0.0746605396270752 # 11:35:29 --- INFO: epoch: 394, batch: 1, loss: 4.13068962097168, modularity_loss: -0.5711579322814941, purity_loss: 4.627171516418457, regularization_loss: 0.07467612624168396 # 11:35:29 --- INFO: epoch: 395, batch: 1, loss: 4.130197048187256, modularity_loss: -0.5709084272384644, purity_loss: 4.626416206359863, regularization_loss: 0.07468926906585693 # 11:35:30 --- INFO: epoch: 396, batch: 1, loss: 4.129792213439941, modularity_loss: -0.5707371234893799, purity_loss: 4.625830173492432, regularization_loss: 0.07469898462295532 # 11:35:30 --- INFO: epoch: 397, batch: 1, loss: 4.129382610321045, modularity_loss: -0.5706547498703003, purity_loss: 4.625332355499268, regularization_loss: 0.0747048631310463 # 11:35:30 --- INFO: epoch: 398, batch: 1, loss: 4.128902912139893, modularity_loss: -0.5706552267074585, purity_loss: 4.624850749969482, regularization_loss: 0.0747072622179985 # 11:35:30 --- INFO: epoch: 399, batch: 1, loss: 4.128324508666992, modularity_loss: -0.5707212686538696, purity_loss: 4.624338626861572, regularization_loss: 0.0747070461511612 # 11:35:31 --- INFO: epoch: 400, batch: 1, loss: 4.1276631355285645, modularity_loss: -0.5708315968513489, purity_loss: 4.623789310455322, regularization_loss: 0.0747053474187851 # 11:35:31 --- INFO: epoch: 401, batch: 1, loss: 4.126969337463379, modularity_loss: -0.5709680914878845, purity_loss: 4.623234272003174, regularization_loss: 0.07470308244228363 # 11:35:31 --- INFO: epoch: 402, batch: 1, loss: 4.126283645629883, modularity_loss: -0.5711159706115723, purity_loss: 4.622698783874512, regularization_loss: 0.07470092922449112 # 11:35:32 --- INFO: epoch: 403, batch: 1, loss: 4.1256422996521, modularity_loss: -0.5712635517120361, purity_loss: 4.622206687927246, regularization_loss: 0.07469936460256577 # 11:35:32 --- INFO: epoch: 404, batch: 1, loss: 4.125051975250244, modularity_loss: -0.5714000463485718, purity_loss: 4.621753215789795, regularization_loss: 0.07469869405031204 # 11:35:32 --- INFO: epoch: 405, batch: 1, loss: 4.124507427215576, modularity_loss: -0.5715119242668152, purity_loss: 4.6213202476501465, regularization_loss: 0.07469917833805084 # 11:35:33 --- INFO: epoch: 406, batch: 1, loss: 4.124002933502197, modularity_loss: -0.5715882778167725, purity_loss: 4.620890140533447, regularization_loss: 0.07470102608203888 # 11:35:33 --- INFO: epoch: 407, batch: 1, loss: 4.123523235321045, modularity_loss: -0.5716185569763184, purity_loss: 4.6204376220703125, regularization_loss: 0.07470432668924332 # 11:35:33 --- INFO: epoch: 408, batch: 1, loss: 4.123054504394531, modularity_loss: -0.5715982913970947, purity_loss: 4.619944095611572, regularization_loss: 0.07470885664224625 # 11:35:34 --- INFO: epoch: 409, batch: 1, loss: 4.122582912445068, modularity_loss: -0.5715360641479492, purity_loss: 4.619405269622803, regularization_loss: 0.07471392303705215 # 11:35:34 --- INFO: epoch: 410, batch: 1, loss: 4.122095108032227, modularity_loss: -0.5714465975761414, purity_loss: 4.618823051452637, regularization_loss: 0.0747184306383133 # 11:35:34 --- INFO: epoch: 411, batch: 1, loss: 4.12158727645874, modularity_loss: -0.5713443160057068, purity_loss: 4.6182098388671875, regularization_loss: 0.07472154498100281 # 11:35:34 --- INFO: epoch: 412, batch: 1, loss: 4.121058464050293, modularity_loss: -0.5712381601333618, purity_loss: 4.6175737380981445, regularization_loss: 0.0747227892279625 # 11:35:35 --- INFO: epoch: 413, batch: 1, loss: 4.120516300201416, modularity_loss: -0.5711333751678467, purity_loss: 4.616927623748779, regularization_loss: 0.07472220808267593 # 11:35:35 --- INFO: epoch: 414, batch: 1, loss: 4.119972229003906, modularity_loss: -0.5710283517837524, purity_loss: 4.6162800788879395, regularization_loss: 0.07472027093172073 # 11:35:35 --- INFO: epoch: 415, batch: 1, loss: 4.1194376945495605, modularity_loss: -0.5709209442138672, purity_loss: 4.615641117095947, regularization_loss: 0.07471771538257599 # 11:35:36 --- INFO: epoch: 416, batch: 1, loss: 4.118921279907227, modularity_loss: -0.5708101391792297, purity_loss: 4.615016460418701, regularization_loss: 0.07471514493227005 # 11:35:36 --- INFO: epoch: 417, batch: 1, loss: 4.1184210777282715, modularity_loss: -0.5706971883773804, purity_loss: 4.614405155181885, regularization_loss: 0.07471306622028351 # 11:35:36 --- INFO: epoch: 418, batch: 1, loss: 4.117928504943848, modularity_loss: -0.5705853700637817, purity_loss: 4.613802433013916, regularization_loss: 0.07471168786287308 # 11:35:37 --- INFO: epoch: 419, batch: 1, loss: 4.117438793182373, modularity_loss: -0.5704784393310547, purity_loss: 4.613206386566162, regularization_loss: 0.07471103221178055 # 11:35:37 --- INFO: epoch: 420, batch: 1, loss: 4.116943359375, modularity_loss: -0.5703818202018738, purity_loss: 4.612614154815674, regularization_loss: 0.0747108981013298 # 11:35:37 --- INFO: epoch: 421, batch: 1, loss: 4.116428375244141, modularity_loss: -0.5702944993972778, purity_loss: 4.612011432647705, regularization_loss: 0.07471119612455368 # 11:35:37 --- INFO: epoch: 422, batch: 1, loss: 4.1158976554870605, modularity_loss: -0.5702118873596191, purity_loss: 4.611397743225098, regularization_loss: 0.07471181452274323 # 11:35:38 --- INFO: epoch: 423, batch: 1, loss: 4.11535120010376, modularity_loss: -0.57012939453125, purity_loss: 4.610767841339111, regularization_loss: 0.07471265643835068 # 11:35:38 --- INFO: epoch: 424, batch: 1, loss: 4.114794731140137, modularity_loss: -0.5700443387031555, purity_loss: 4.610125541687012, regularization_loss: 0.0747133120894432 # 11:35:38 --- INFO: epoch: 425, batch: 1, loss: 4.114228248596191, modularity_loss: -0.5699565410614014, purity_loss: 4.609471321105957, regularization_loss: 0.07471322268247604 # 11:35:39 --- INFO: epoch: 426, batch: 1, loss: 4.1136474609375, modularity_loss: -0.5698621273040771, purity_loss: 4.608797550201416, regularization_loss: 0.07471185177564621 # 11:35:39 --- INFO: epoch: 427, batch: 1, loss: 4.113059043884277, modularity_loss: -0.5697620511054993, purity_loss: 4.608112335205078, regularization_loss: 0.07470874488353729 # 11:35:39 --- INFO: epoch: 428, batch: 1, loss: 4.112453460693359, modularity_loss: -0.5696607232093811, purity_loss: 4.607410907745361, regularization_loss: 0.07470352947711945 # 11:35:40 --- INFO: epoch: 429, batch: 1, loss: 4.111816883087158, modularity_loss: -0.5695497989654541, purity_loss: 4.60667085647583, regularization_loss: 0.07469603419303894 # 11:35:40 --- INFO: epoch: 430, batch: 1, loss: 4.111144542694092, modularity_loss: -0.5694237947463989, purity_loss: 4.605882167816162, regularization_loss: 0.07468615472316742 # 11:35:40 --- INFO: epoch: 431, batch: 1, loss: 4.110429763793945, modularity_loss: -0.5692689418792725, purity_loss: 4.605024814605713, regularization_loss: 0.0746741071343422 # 11:35:40 --- INFO: epoch: 432, batch: 1, loss: 4.109653949737549, modularity_loss: -0.5690774917602539, purity_loss: 4.604071140289307, regularization_loss: 0.07466008514165878 # 11:35:41 --- INFO: epoch: 433, batch: 1, loss: 4.1088104248046875, modularity_loss: -0.5688424110412598, purity_loss: 4.60300874710083, regularization_loss: 0.07464385777711868 # 11:35:41 --- INFO: epoch: 434, batch: 1, loss: 4.107895851135254, modularity_loss: -0.5685737133026123, purity_loss: 4.601844787597656, regularization_loss: 0.07462464272975922 # 11:35:41 --- INFO: epoch: 435, batch: 1, loss: 4.106884479522705, modularity_loss: -0.5682656168937683, purity_loss: 4.600548267364502, regularization_loss: 0.07460162043571472 # 11:35:42 --- INFO: epoch: 436, batch: 1, loss: 4.105752468109131, modularity_loss: -0.5679073333740234, purity_loss: 4.599085807800293, regularization_loss: 0.07457396388053894 # 11:35:42 --- INFO: epoch: 437, batch: 1, loss: 4.10446834564209, modularity_loss: -0.5674849152565002, purity_loss: 4.597412109375, regularization_loss: 0.0745411366224289 # 11:35:42 --- INFO: epoch: 438, batch: 1, loss: 4.102963924407959, modularity_loss: -0.5669894218444824, purity_loss: 4.595451831817627, regularization_loss: 0.07450137287378311 # 11:35:43 --- INFO: epoch: 439, batch: 1, loss: 4.101215839385986, modularity_loss: -0.5664106607437134, purity_loss: 4.593174457550049, regularization_loss: 0.07445225864648819 # 11:35:43 --- INFO: epoch: 440, batch: 1, loss: 4.099147319793701, modularity_loss: -0.5657711029052734, purity_loss: 4.590527057647705, regularization_loss: 0.07439125329256058 # 11:35:43 --- INFO: epoch: 441, batch: 1, loss: 4.096593856811523, modularity_loss: -0.5650631189346313, purity_loss: 4.587341785430908, regularization_loss: 0.07431499660015106 # 11:35:43 --- INFO: epoch: 442, batch: 1, loss: 4.093234539031982, modularity_loss: -0.5642626285552979, purity_loss: 4.583278656005859, regularization_loss: 0.07421834021806717 # 11:35:44 --- INFO: epoch: 443, batch: 1, loss: 4.088987827301025, modularity_loss: -0.5634028911590576, purity_loss: 4.578296661376953, regularization_loss: 0.0740942433476448 # 11:35:44 --- INFO: epoch: 444, batch: 1, loss: 4.083475112915039, modularity_loss: -0.5625929236412048, purity_loss: 4.57213020324707, regularization_loss: 0.07393770664930344 # 11:35:44 --- INFO: epoch: 445, batch: 1, loss: 4.076183795928955, modularity_loss: -0.5620026588439941, purity_loss: 4.564441204071045, regularization_loss: 0.07374510914087296 # 11:35:45 --- INFO: epoch: 446, batch: 1, loss: 4.066742420196533, modularity_loss: -0.5619003176689148, purity_loss: 4.555125713348389, regularization_loss: 0.07351697236299515 # 11:35:45 --- INFO: epoch: 447, batch: 1, loss: 4.055968761444092, modularity_loss: -0.5626815557479858, purity_loss: 4.545380115509033, regularization_loss: 0.07327021658420563 # 11:35:45 --- INFO: epoch: 448, batch: 1, loss: 4.045150279998779, modularity_loss: -0.5646786689758301, purity_loss: 4.536784648895264, regularization_loss: 0.07304443418979645 # 11:35:46 --- INFO: epoch: 449, batch: 1, loss: 4.0336594581604, modularity_loss: -0.5677903890609741, purity_loss: 4.528567314147949, regularization_loss: 0.07288264483213425 # 11:35:46 --- INFO: epoch: 450, batch: 1, loss: 4.018993854522705, modularity_loss: -0.5715136528015137, purity_loss: 4.517703056335449, regularization_loss: 0.0728045403957367 # 11:35:46 --- INFO: epoch: 451, batch: 1, loss: 4.001464366912842, modularity_loss: -0.5752850770950317, purity_loss: 4.503946304321289, regularization_loss: 0.07280334085226059 # 11:35:47 --- INFO: epoch: 452, batch: 1, loss: 3.9833433628082275, modularity_loss: -0.5787436962127686, purity_loss: 4.489232063293457, regularization_loss: 0.07285504788160324 # 11:35:47 --- INFO: epoch: 453, batch: 1, loss: 3.965441942214966, modularity_loss: -0.5816754698753357, purity_loss: 4.474185943603516, regularization_loss: 0.07293146103620529 # 11:35:47 --- INFO: epoch: 454, batch: 1, loss: 3.9478325843811035, modularity_loss: -0.5840556621551514, purity_loss: 4.458881378173828, regularization_loss: 0.07300682365894318 # 11:35:47 --- INFO: epoch: 455, batch: 1, loss: 3.929542064666748, modularity_loss: -0.5858882665634155, purity_loss: 4.442365646362305, regularization_loss: 0.07306475937366486 # 11:35:48 --- INFO: epoch: 456, batch: 1, loss: 3.910885810852051, modularity_loss: -0.5873255729675293, purity_loss: 4.425100803375244, regularization_loss: 0.0731106549501419 # 11:35:48 --- INFO: epoch: 457, batch: 1, loss: 3.8936257362365723, modularity_loss: -0.5885381698608398, purity_loss: 4.409008979797363, regularization_loss: 0.07315487414598465 # 11:35:48 --- INFO: epoch: 458, batch: 1, loss: 3.878166437149048, modularity_loss: -0.5897192358970642, purity_loss: 4.394680500030518, regularization_loss: 0.07320515811443329 # 11:35:49 --- INFO: epoch: 459, batch: 1, loss: 3.8634755611419678, modularity_loss: -0.5910282135009766, purity_loss: 4.381239414215088, regularization_loss: 0.07326427847146988 # 11:35:49 --- INFO: epoch: 460, batch: 1, loss: 3.848785638809204, modularity_loss: -0.5925470590591431, purity_loss: 4.368001461029053, regularization_loss: 0.07333125919103622 # 11:35:49 --- INFO: epoch: 461, batch: 1, loss: 3.8346896171569824, modularity_loss: -0.5942522287368774, purity_loss: 4.355537414550781, regularization_loss: 0.073404461145401 # 11:35:50 --- INFO: epoch: 462, batch: 1, loss: 3.8222479820251465, modularity_loss: -0.5959993600845337, purity_loss: 4.344768524169922, regularization_loss: 0.07347877323627472 # 11:35:50 --- INFO: epoch: 463, batch: 1, loss: 3.810962438583374, modularity_loss: -0.5976624488830566, purity_loss: 4.3350725173950195, regularization_loss: 0.07355231046676636 # 11:35:50 --- INFO: epoch: 464, batch: 1, loss: 3.800318717956543, modularity_loss: -0.599134624004364, purity_loss: 4.325829982757568, regularization_loss: 0.07362334430217743 # 11:35:50 --- INFO: epoch: 465, batch: 1, loss: 3.789701461791992, modularity_loss: -0.6004000902175903, purity_loss: 4.316411018371582, regularization_loss: 0.0736904889345169 # 11:35:51 --- INFO: epoch: 466, batch: 1, loss: 3.779043436050415, modularity_loss: -0.6014965772628784, purity_loss: 4.306787014007568, regularization_loss: 0.07375302165746689 # 11:35:51 --- INFO: epoch: 467, batch: 1, loss: 3.7683043479919434, modularity_loss: -0.6024739742279053, purity_loss: 4.29696798324585, regularization_loss: 0.07381034642457962 # 11:35:51 --- INFO: epoch: 468, batch: 1, loss: 3.7576615810394287, modularity_loss: -0.6033977270126343, purity_loss: 4.287195205688477, regularization_loss: 0.07386408746242523 # 11:35:52 --- INFO: epoch: 469, batch: 1, loss: 3.7471375465393066, modularity_loss: -0.6043241024017334, purity_loss: 4.277544021606445, regularization_loss: 0.07391761988401413 # 11:35:52 --- INFO: epoch: 470, batch: 1, loss: 3.736737012863159, modularity_loss: -0.6052950024604797, purity_loss: 4.268056869506836, regularization_loss: 0.07397517561912537 # 11:35:52 --- INFO: epoch: 471, batch: 1, loss: 3.726527452468872, modularity_loss: -0.6063229441642761, purity_loss: 4.258810520172119, regularization_loss: 0.07403992116451263 # 11:35:53 --- INFO: epoch: 472, batch: 1, loss: 3.716978073120117, modularity_loss: -0.6073740124702454, purity_loss: 4.250239849090576, regularization_loss: 0.07411223649978638 # 11:35:53 --- INFO: epoch: 473, batch: 1, loss: 3.708373546600342, modularity_loss: -0.6083994507789612, purity_loss: 4.242583274841309, regularization_loss: 0.07418975979089737 # 11:35:53 --- INFO: epoch: 474, batch: 1, loss: 3.7009212970733643, modularity_loss: -0.6093660593032837, purity_loss: 4.23601770401001, regularization_loss: 0.07426965236663818 # 11:35:53 --- INFO: epoch: 475, batch: 1, loss: 3.6947288513183594, modularity_loss: -0.6102364659309387, purity_loss: 4.230618953704834, regularization_loss: 0.07434624433517456 # 11:35:54 --- INFO: epoch: 476, batch: 1, loss: 3.6895663738250732, modularity_loss: -0.6109839677810669, purity_loss: 4.226137161254883, regularization_loss: 0.07441326230764389 # 11:35:54 --- INFO: epoch: 477, batch: 1, loss: 3.6851806640625, modularity_loss: -0.6116119027137756, purity_loss: 4.222325801849365, regularization_loss: 0.07446686178445816 # 11:35:54 --- INFO: epoch: 478, batch: 1, loss: 3.6812422275543213, modularity_loss: -0.6121276021003723, purity_loss: 4.218865394592285, regularization_loss: 0.07450444996356964 # 11:35:55 --- INFO: epoch: 479, batch: 1, loss: 3.677539825439453, modularity_loss: -0.6125479936599731, purity_loss: 4.215561389923096, regularization_loss: 0.07452645897865295 # 11:35:55 --- INFO: epoch: 480, batch: 1, loss: 3.673973560333252, modularity_loss: -0.6128899455070496, purity_loss: 4.2123284339904785, regularization_loss: 0.07453514635562897 # 11:35:55 --- INFO: epoch: 481, batch: 1, loss: 3.670515775680542, modularity_loss: -0.6131739020347595, purity_loss: 4.209155082702637, regularization_loss: 0.07453455775976181 # 11:35:56 --- INFO: epoch: 482, batch: 1, loss: 3.667203426361084, modularity_loss: -0.6134227514266968, purity_loss: 4.20609712600708, regularization_loss: 0.07452902942895889 # 11:35:56 --- INFO: epoch: 483, batch: 1, loss: 3.6641271114349365, modularity_loss: -0.6136577129364014, purity_loss: 4.2032623291015625, regularization_loss: 0.07452241331338882 # 11:35:56 --- INFO: epoch: 484, batch: 1, loss: 3.6613268852233887, modularity_loss: -0.6138933897018433, purity_loss: 4.200702667236328, regularization_loss: 0.07451746612787247 # 11:35:56 --- INFO: epoch: 485, batch: 1, loss: 3.6587977409362793, modularity_loss: -0.614139199256897, purity_loss: 4.198421001434326, regularization_loss: 0.0745159313082695 # 11:35:57 --- INFO: epoch: 486, batch: 1, loss: 3.6564364433288574, modularity_loss: -0.6143894195556641, purity_loss: 4.19630765914917, regularization_loss: 0.07451830059289932 # 11:35:57 --- INFO: epoch: 487, batch: 1, loss: 3.654141426086426, modularity_loss: -0.6146302223205566, purity_loss: 4.194247245788574, regularization_loss: 0.07452435791492462 # 11:35:57 --- INFO: epoch: 488, batch: 1, loss: 3.651805877685547, modularity_loss: -0.614840567111969, purity_loss: 4.192112922668457, regularization_loss: 0.07453348487615585 # 11:35:58 --- INFO: epoch: 489, batch: 1, loss: 3.6493313312530518, modularity_loss: -0.6149961948394775, purity_loss: 4.189782619476318, regularization_loss: 0.07454492151737213 # 11:35:58 --- INFO: epoch: 490, batch: 1, loss: 3.646756172180176, modularity_loss: -0.6150829195976257, purity_loss: 4.187281131744385, regularization_loss: 0.07455798238515854 # 11:35:58 --- INFO: epoch: 491, batch: 1, loss: 3.644167423248291, modularity_loss: -0.6150844097137451, purity_loss: 4.184680461883545, regularization_loss: 0.07457150518894196 # 11:35:59 --- INFO: epoch: 492, batch: 1, loss: 3.6415398120880127, modularity_loss: -0.6149938106536865, purity_loss: 4.181949615478516, regularization_loss: 0.07458393275737762 # 11:35:59 --- INFO: epoch: 493, batch: 1, loss: 3.638885498046875, modularity_loss: -0.6148231029510498, purity_loss: 4.17911434173584, regularization_loss: 0.0745941624045372 # 11:35:59 --- INFO: epoch: 494, batch: 1, loss: 3.636228561401367, modularity_loss: -0.6145913004875183, purity_loss: 4.1762189865112305, regularization_loss: 0.07460074126720428 # 11:36:00 --- INFO: epoch: 495, batch: 1, loss: 3.6336047649383545, modularity_loss: -0.6143289804458618, purity_loss: 4.173331260681152, regularization_loss: 0.074602410197258 # 11:36:00 --- INFO: epoch: 496, batch: 1, loss: 3.631000280380249, modularity_loss: -0.6140619516372681, purity_loss: 4.170464515686035, regularization_loss: 0.07459764927625656 # 11:36:00 --- INFO: epoch: 497, batch: 1, loss: 3.6283230781555176, modularity_loss: -0.6138187050819397, purity_loss: 4.167555809020996, regularization_loss: 0.0745859295129776 # 11:36:01 --- INFO: epoch: 498, batch: 1, loss: 3.625584602355957, modularity_loss: -0.6136263608932495, purity_loss: 4.164642810821533, regularization_loss: 0.07456820458173752 # 11:36:01 --- INFO: epoch: 499, batch: 1, loss: 3.622783660888672, modularity_loss: -0.6134976744651794, purity_loss: 4.1617350578308105, regularization_loss: 0.07454626262187958 # 11:36:01 --- INFO: epoch: 500, batch: 1, loss: 3.6199400424957275, modularity_loss: -0.613420844078064, purity_loss: 4.158838748931885, regularization_loss: 0.07452220469713211 # 11:36:02 --- INFO: epoch: 501, batch: 1, loss: 3.6171023845672607, modularity_loss: -0.6133748888969421, purity_loss: 4.155978679656982, regularization_loss: 0.07449851930141449 # 11:36:02 --- INFO: epoch: 502, batch: 1, loss: 3.614269733428955, modularity_loss: -0.6133327484130859, purity_loss: 4.153124809265137, regularization_loss: 0.07447752356529236 # 11:36:02 --- INFO: epoch: 503, batch: 1, loss: 3.611417531967163, modularity_loss: -0.6132693290710449, purity_loss: 4.15022611618042, regularization_loss: 0.07446078211069107 # 11:36:03 --- INFO: epoch: 504, batch: 1, loss: 3.608511209487915, modularity_loss: -0.613166868686676, purity_loss: 4.147229194641113, regularization_loss: 0.07444890588521957 # 11:36:03 --- INFO: epoch: 505, batch: 1, loss: 3.605525016784668, modularity_loss: -0.6130160093307495, purity_loss: 4.144099235534668, regularization_loss: 0.07444173097610474 # 11:36:03 --- INFO: epoch: 506, batch: 1, loss: 3.6024580001831055, modularity_loss: -0.6128139495849609, purity_loss: 4.140833854675293, regularization_loss: 0.07443820685148239 # 11:36:04 --- INFO: epoch: 507, batch: 1, loss: 3.5993258953094482, modularity_loss: -0.6125696897506714, purity_loss: 4.137458324432373, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 508, batch: 1, loss: 3.5961737632751465, modularity_loss: -0.6122933030128479, purity_loss: 4.134029865264893, regularization_loss: 0.07443727552890778 # 11:36:04 --- INFO: epoch: 509, batch: 1, loss: 3.5930728912353516, modularity_loss: -0.6120021939277649, purity_loss: 4.130638122558594, regularization_loss: 0.07443708181381226 # 11:36:05 --- INFO: epoch: 510, batch: 1, loss: 3.590001106262207, modularity_loss: -0.611706554889679, purity_loss: 4.127272605895996, regularization_loss: 0.07443508505821228 # 11:36:05 --- INFO: epoch: 511, batch: 1, loss: 3.5869479179382324, modularity_loss: -0.6114233732223511, purity_loss: 4.123940944671631, regularization_loss: 0.0744304358959198 # 11:36:05 --- INFO: epoch: 512, batch: 1, loss: 3.5838894844055176, modularity_loss: -0.6111712455749512, purity_loss: 4.1206374168396, regularization_loss: 0.07442330569028854 # 11:36:06 --- INFO: epoch: 513, batch: 1, loss: 3.580822229385376, modularity_loss: -0.6109635829925537, purity_loss: 4.117371559143066, regularization_loss: 0.07441431283950806 # 11:36:06 --- INFO: epoch: 514, batch: 1, loss: 3.577752113342285, modularity_loss: -0.6107942461967468, purity_loss: 4.114143371582031, regularization_loss: 0.07440303266048431 # 11:36:06 --- INFO: epoch: 515, batch: 1, loss: 3.574666976928711, modularity_loss: -0.6106579303741455, purity_loss: 4.110934734344482, regularization_loss: 0.0743902325630188 # 11:36:06 --- INFO: epoch: 516, batch: 1, loss: 3.571580648422241, modularity_loss: -0.6105481386184692, purity_loss: 4.107751846313477, regularization_loss: 0.07437692582607269 # 11:36:07 --- INFO: epoch: 517, batch: 1, loss: 3.568519115447998, modularity_loss: -0.6104516983032227, purity_loss: 4.104607105255127, regularization_loss: 0.07436370104551315 # 11:36:07 --- INFO: epoch: 518, batch: 1, loss: 3.565471649169922, modularity_loss: -0.6103577613830566, purity_loss: 4.101478099822998, regularization_loss: 0.07435141503810883 # 11:36:07 --- INFO: epoch: 519, batch: 1, loss: 3.562424421310425, modularity_loss: -0.610254168510437, purity_loss: 4.0983381271362305, regularization_loss: 0.07434046268463135 # 11:36:08 --- INFO: epoch: 520, batch: 1, loss: 3.5593724250793457, modularity_loss: -0.6101377010345459, purity_loss: 4.095178604125977, regularization_loss: 0.07433146238327026 # 11:36:08 --- INFO: epoch: 521, batch: 1, loss: 3.556313991546631, modularity_loss: -0.6100000143051147, purity_loss: 4.091989994049072, regularization_loss: 0.0743240937590599 # 11:36:09 --- INFO: epoch: 522, batch: 1, loss: 3.553253412246704, modularity_loss: -0.6098423004150391, purity_loss: 4.088777542114258, regularization_loss: 0.07431814074516296 # 11:36:09 --- INFO: epoch: 523, batch: 1, loss: 3.5502114295959473, modularity_loss: -0.6096738576889038, purity_loss: 4.0855712890625, regularization_loss: 0.07431399822235107 # 11:36:09 --- INFO: epoch: 524, batch: 1, loss: 3.5471713542938232, modularity_loss: -0.6095079183578491, purity_loss: 4.082367420196533, regularization_loss: 0.07431186735630035 # 11:36:10 --- INFO: epoch: 525, batch: 1, loss: 3.544124126434326, modularity_loss: -0.6093576550483704, purity_loss: 4.079169750213623, regularization_loss: 0.07431215792894363 # 11:36:10 --- INFO: epoch: 526, batch: 1, loss: 3.5410468578338623, modularity_loss: -0.6092305779457092, purity_loss: 4.075963020324707, regularization_loss: 0.07431443780660629 # 11:36:10 --- INFO: epoch: 527, batch: 1, loss: 3.5379300117492676, modularity_loss: -0.6091315150260925, purity_loss: 4.072742938995361, regularization_loss: 0.07431862503290176 # 11:36:11 --- INFO: epoch: 528, batch: 1, loss: 3.53481125831604, modularity_loss: -0.6090551018714905, purity_loss: 4.069542407989502, regularization_loss: 0.07432398945093155 # 11:36:11 --- INFO: epoch: 529, batch: 1, loss: 3.5316741466522217, modularity_loss: -0.6089891195297241, purity_loss: 4.066333770751953, regularization_loss: 0.07432956993579865 # 11:36:11 --- INFO: epoch: 530, batch: 1, loss: 3.5285391807556152, modularity_loss: -0.6089223623275757, purity_loss: 4.063127040863037, regularization_loss: 0.07433458417654037 # 11:36:12 --- INFO: epoch: 531, batch: 1, loss: 3.5254108905792236, modularity_loss: -0.6088444590568542, purity_loss: 4.059917449951172, regularization_loss: 0.07433795928955078 # 11:36:12 --- INFO: epoch: 532, batch: 1, loss: 3.522289991378784, modularity_loss: -0.6087523698806763, purity_loss: 4.056702613830566, regularization_loss: 0.07433980703353882 # 11:36:12 --- INFO: epoch: 533, batch: 1, loss: 3.5191810131073, modularity_loss: -0.6086558699607849, purity_loss: 4.053495407104492, regularization_loss: 0.07434152811765671 # 11:36:13 --- INFO: epoch: 534, batch: 1, loss: 3.5160837173461914, modularity_loss: -0.6085688471794128, purity_loss: 4.050307750701904, regularization_loss: 0.07434485107660294 # 11:36:13 --- INFO: epoch: 535, batch: 1, loss: 3.5129997730255127, modularity_loss: -0.6084942817687988, purity_loss: 4.047143459320068, regularization_loss: 0.07435066252946854 # 11:36:13 --- INFO: epoch: 536, batch: 1, loss: 3.509936809539795, modularity_loss: -0.6084328293800354, purity_loss: 4.044010639190674, regularization_loss: 0.07435906678438187 # 11:36:14 --- INFO: epoch: 537, batch: 1, loss: 3.5069172382354736, modularity_loss: -0.6083766222000122, purity_loss: 4.040925025939941, regularization_loss: 0.074368916451931 # 11:36:14 --- INFO: epoch: 538, batch: 1, loss: 3.5039377212524414, modularity_loss: -0.608316957950592, purity_loss: 4.037875652313232, regularization_loss: 0.07437888532876968 # 11:36:14 --- INFO: epoch: 539, batch: 1, loss: 3.5009870529174805, modularity_loss: -0.6082561016082764, purity_loss: 4.034853935241699, regularization_loss: 0.07438908517360687 # 11:36:15 --- INFO: epoch: 540, batch: 1, loss: 3.4980661869049072, modularity_loss: -0.608196496963501, purity_loss: 4.031863212585449, regularization_loss: 0.07439954578876495 # 11:36:15 --- INFO: epoch: 541, batch: 1, loss: 3.495177745819092, modularity_loss: -0.6081417202949524, purity_loss: 4.028908729553223, regularization_loss: 0.0744108110666275 # 11:36:16 --- INFO: epoch: 542, batch: 1, loss: 3.4923415184020996, modularity_loss: -0.608096718788147, purity_loss: 4.02601432800293, regularization_loss: 0.07442396879196167 # 11:36:16 --- INFO: epoch: 543, batch: 1, loss: 3.489562511444092, modularity_loss: -0.6080563068389893, purity_loss: 4.02318000793457, regularization_loss: 0.07443877309560776 # 11:36:17 --- INFO: epoch: 544, batch: 1, loss: 3.4868764877319336, modularity_loss: -0.6080097556114197, purity_loss: 4.020432472229004, regularization_loss: 0.07445371150970459 # 11:36:17 --- INFO: epoch: 545, batch: 1, loss: 3.4843056201934814, modularity_loss: -0.607948899269104, purity_loss: 4.017787456512451, regularization_loss: 0.07446698099374771 # 11:36:17 --- INFO: epoch: 546, batch: 1, loss: 3.481825828552246, modularity_loss: -0.6078734993934631, purity_loss: 4.015221118927002, regularization_loss: 0.07447806745767593 # 11:36:18 --- INFO: epoch: 547, batch: 1, loss: 3.479466438293457, modularity_loss: -0.6077962517738342, purity_loss: 4.012774467468262, regularization_loss: 0.07448829710483551 # 11:36:18 --- INFO: epoch: 548, batch: 1, loss: 3.4772183895111084, modularity_loss: -0.6077297925949097, purity_loss: 4.010449409484863, regularization_loss: 0.07449881732463837 # 11:36:18 --- INFO: epoch: 549, batch: 1, loss: 3.475069999694824, modularity_loss: -0.6076837778091431, purity_loss: 4.008243083953857, regularization_loss: 0.07451068609952927 # 11:36:19 --- INFO: epoch: 550, batch: 1, loss: 3.4730210304260254, modularity_loss: -0.6076596975326538, purity_loss: 4.006156921386719, regularization_loss: 0.07452377676963806 # 11:36:19 --- INFO: epoch: 551, batch: 1, loss: 3.471071243286133, modularity_loss: -0.6076537370681763, purity_loss: 4.004188060760498, regularization_loss: 0.07453705370426178 # 11:36:19 --- INFO: epoch: 552, batch: 1, loss: 3.4692130088806152, modularity_loss: -0.607658326625824, purity_loss: 4.002322196960449, regularization_loss: 0.07454905658960342 # 11:36:20 --- INFO: epoch: 553, batch: 1, loss: 3.4674367904663086, modularity_loss: -0.6076701879501343, purity_loss: 4.000547409057617, regularization_loss: 0.07455962151288986 # 11:36:20 --- INFO: epoch: 554, batch: 1, loss: 3.465729236602783, modularity_loss: -0.6076894998550415, purity_loss: 3.9988491535186768, regularization_loss: 0.0745697021484375 # 11:36:20 --- INFO: epoch: 555, batch: 1, loss: 3.464085578918457, modularity_loss: -0.6077148914337158, purity_loss: 3.997220516204834, regularization_loss: 0.07457991689443588 # 11:36:21 --- INFO: epoch: 556, batch: 1, loss: 3.4624972343444824, modularity_loss: -0.6077462434768677, purity_loss: 3.995652914047241, regularization_loss: 0.0745905190706253 # 11:36:21 --- INFO: epoch: 557, batch: 1, loss: 3.460960626602173, modularity_loss: -0.6077839136123657, purity_loss: 3.9941442012786865, regularization_loss: 0.07460035383701324 # 11:36:22 --- INFO: epoch: 558, batch: 1, loss: 3.459486722946167, modularity_loss: -0.6078293323516846, purity_loss: 3.9927077293395996, regularization_loss: 0.07460832595825195 # 11:36:22 --- INFO: epoch: 559, batch: 1, loss: 3.4580631256103516, modularity_loss: -0.6078857779502869, purity_loss: 3.9913344383239746, regularization_loss: 0.0746145099401474 # 11:36:22 --- INFO: epoch: 560, batch: 1, loss: 3.4566872119903564, modularity_loss: -0.6079564094543457, purity_loss: 3.9900238513946533, regularization_loss: 0.07461968809366226 # 11:36:23 --- INFO: epoch: 561, batch: 1, loss: 3.4553627967834473, modularity_loss: -0.6080377101898193, purity_loss: 3.9887757301330566, regularization_loss: 0.07462484389543533 # 11:36:23 --- INFO: epoch: 562, batch: 1, loss: 3.454085350036621, modularity_loss: -0.608126163482666, purity_loss: 3.9875810146331787, regularization_loss: 0.07463043183088303 # 11:36:23 --- INFO: epoch: 563, batch: 1, loss: 3.4528565406799316, modularity_loss: -0.6082156300544739, purity_loss: 3.986436128616333, regularization_loss: 0.07463616132736206 # 11:36:24 --- INFO: epoch: 564, batch: 1, loss: 3.451673984527588, modularity_loss: -0.6083011627197266, purity_loss: 3.9853339195251465, regularization_loss: 0.07464126497507095 # 11:36:24 --- INFO: epoch: 565, batch: 1, loss: 3.450528621673584, modularity_loss: -0.6083787679672241, purity_loss: 3.984261989593506, regularization_loss: 0.07464525103569031 # 11:36:24 --- INFO: epoch: 566, batch: 1, loss: 3.44942307472229, modularity_loss: -0.6084476113319397, purity_loss: 3.983222246170044, regularization_loss: 0.07464844733476639 # 11:36:25 --- INFO: epoch: 567, batch: 1, loss: 3.448355197906494, modularity_loss: -0.6085067987442017, purity_loss: 3.982210636138916, regularization_loss: 0.07465138286352158 # 11:36:25 --- INFO: epoch: 568, batch: 1, loss: 3.447329044342041, modularity_loss: -0.6085564494132996, purity_loss: 3.981231212615967, regularization_loss: 0.07465439289808273 # 11:36:25 --- INFO: epoch: 569, batch: 1, loss: 3.4463324546813965, modularity_loss: -0.6085982322692871, purity_loss: 3.980273485183716, regularization_loss: 0.07465726137161255 # 11:36:26 --- INFO: epoch: 570, batch: 1, loss: 3.4453659057617188, modularity_loss: -0.6086323857307434, purity_loss: 3.9793386459350586, regularization_loss: 0.07465960830450058 # 11:36:26 --- INFO: epoch: 571, batch: 1, loss: 3.444427490234375, modularity_loss: -0.6086603403091431, purity_loss: 3.978426456451416, regularization_loss: 0.07466133683919907 # 11:36:26 --- INFO: epoch: 572, batch: 1, loss: 3.443511486053467, modularity_loss: -0.6086817979812622, purity_loss: 3.9775304794311523, regularization_loss: 0.07466293126344681 # 11:36:27 --- INFO: epoch: 573, batch: 1, loss: 3.44262433052063, modularity_loss: -0.6086944341659546, purity_loss: 3.976653575897217, regularization_loss: 0.0746651366353035 # 11:36:27 --- INFO: epoch: 574, batch: 1, loss: 3.441760540008545, modularity_loss: -0.6086971759796143, purity_loss: 3.9757895469665527, regularization_loss: 0.07466808706521988 # 11:36:27 --- INFO: epoch: 575, batch: 1, loss: 3.4409244060516357, modularity_loss: -0.6086897850036621, purity_loss: 3.974942922592163, regularization_loss: 0.0746712014079094 # 11:36:28 --- INFO: epoch: 576, batch: 1, loss: 3.4401187896728516, modularity_loss: -0.6086709499359131, purity_loss: 3.974116325378418, regularization_loss: 0.07467330247163773 # 11:36:28 --- INFO: epoch: 577, batch: 1, loss: 3.439330577850342, modularity_loss: -0.6086453199386597, purity_loss: 3.973301410675049, regularization_loss: 0.07467443495988846 # 11:36:28 --- INFO: epoch: 578, batch: 1, loss: 3.438565731048584, modularity_loss: -0.6086165904998779, purity_loss: 3.9725069999694824, regularization_loss: 0.07467517256736755 # 11:36:28 --- INFO: epoch: 579, batch: 1, loss: 3.4378228187561035, modularity_loss: -0.6085877418518066, purity_loss: 3.9717342853546143, regularization_loss: 0.07467612624168396 # 11:36:29 --- INFO: epoch: 580, batch: 1, loss: 3.437100410461426, modularity_loss: -0.6085619926452637, purity_loss: 3.970985174179077, regularization_loss: 0.07467734068632126 # 11:36:29 --- INFO: epoch: 581, batch: 1, loss: 3.436394453048706, modularity_loss: -0.608540415763855, purity_loss: 3.9702563285827637, regularization_loss: 0.07467859983444214 # 11:36:29 --- INFO: epoch: 582, batch: 1, loss: 3.4357075691223145, modularity_loss: -0.608521044254303, purity_loss: 3.9695487022399902, regularization_loss: 0.0746799185872078 # 11:36:30 --- INFO: epoch: 583, batch: 1, loss: 3.435039758682251, modularity_loss: -0.6085017919540405, purity_loss: 3.968859910964966, regularization_loss: 0.07468163222074509 # 11:36:30 --- INFO: epoch: 584, batch: 1, loss: 3.4343841075897217, modularity_loss: -0.6084802150726318, purity_loss: 3.9681804180145264, regularization_loss: 0.07468390464782715 # 11:36:30 --- INFO: epoch: 585, batch: 1, loss: 3.4337472915649414, modularity_loss: -0.6084542870521545, purity_loss: 3.967515468597412, regularization_loss: 0.0746862068772316 # 11:36:31 --- INFO: epoch: 586, batch: 1, loss: 3.433120012283325, modularity_loss: -0.6084240078926086, purity_loss: 3.9668564796447754, regularization_loss: 0.07468755543231964 # 11:36:31 --- INFO: epoch: 587, batch: 1, loss: 3.4325149059295654, modularity_loss: -0.6083940267562866, purity_loss: 3.9662208557128906, regularization_loss: 0.07468804717063904 # 11:36:31 --- INFO: epoch: 588, batch: 1, loss: 3.431920051574707, modularity_loss: -0.6083695888519287, purity_loss: 3.965601682662964, regularization_loss: 0.07468798011541367 # 11:36:32 --- INFO: epoch: 589, batch: 1, loss: 3.4313364028930664, modularity_loss: -0.6083557605743408, purity_loss: 3.9650044441223145, regularization_loss: 0.07468771934509277 # 11:36:32 --- INFO: epoch: 590, batch: 1, loss: 3.430765151977539, modularity_loss: -0.6083537340164185, purity_loss: 3.9644315242767334, regularization_loss: 0.07468744367361069 # 11:36:32 --- INFO: epoch: 591, batch: 1, loss: 3.430205821990967, modularity_loss: -0.608362078666687, purity_loss: 3.9638805389404297, regularization_loss: 0.0746874138712883 # 11:36:33 --- INFO: epoch: 592, batch: 1, loss: 3.429654359817505, modularity_loss: -0.6083762049674988, purity_loss: 3.9633426666259766, regularization_loss: 0.07468793541193008 # 11:36:33 --- INFO: epoch: 593, batch: 1, loss: 3.429117202758789, modularity_loss: -0.6083896160125732, purity_loss: 3.962817430496216, regularization_loss: 0.0746893361210823 # 11:36:33 --- INFO: epoch: 594, batch: 1, loss: 3.4285888671875, modularity_loss: -0.6083987355232239, purity_loss: 3.9622962474823, regularization_loss: 0.07469140738248825 # 11:36:33 --- INFO: epoch: 595, batch: 1, loss: 3.428069829940796, modularity_loss: -0.6084021925926208, purity_loss: 3.961778402328491, regularization_loss: 0.07469359785318375 # 11:36:34 --- INFO: epoch: 596, batch: 1, loss: 3.427558422088623, modularity_loss: -0.6084009408950806, purity_loss: 3.96126389503479, regularization_loss: 0.07469537109136581 # 11:36:34 --- INFO: epoch: 597, batch: 1, loss: 3.4270524978637695, modularity_loss: -0.6083990335464478, purity_loss: 3.9607551097869873, regularization_loss: 0.07469645142555237 # 11:36:34 --- INFO: epoch: 598, batch: 1, loss: 3.4265573024749756, modularity_loss: -0.6083987951278687, purity_loss: 3.960259199142456, regularization_loss: 0.07469692826271057 # 11:36:35 --- INFO: epoch: 599, batch: 1, loss: 3.42606782913208, modularity_loss: -0.6084031462669373, purity_loss: 3.9597742557525635, regularization_loss: 0.07469679415225983 # 11:36:35 --- INFO: epoch: 600, batch: 1, loss: 3.425586700439453, modularity_loss: -0.6084128022193909, purity_loss: 3.9593029022216797, regularization_loss: 0.07469645142555237 # 11:36:35 --- INFO: epoch: 601, batch: 1, loss: 3.425116539001465, modularity_loss: -0.6084266901016235, purity_loss: 3.9588470458984375, regularization_loss: 0.07469626516103745 # 11:36:36 --- INFO: epoch: 602, batch: 1, loss: 3.4246468544006348, modularity_loss: -0.6084423065185547, purity_loss: 3.95839262008667, regularization_loss: 0.07469648867845535 # 11:36:36 --- INFO: epoch: 603, batch: 1, loss: 3.424189805984497, modularity_loss: -0.6084585189819336, purity_loss: 3.957951068878174, regularization_loss: 0.0746973305940628 # 11:36:36 --- INFO: epoch: 604, batch: 1, loss: 3.423737049102783, modularity_loss: -0.6084734201431274, purity_loss: 3.9575119018554688, regularization_loss: 0.07469865679740906 # 11:36:36 --- INFO: epoch: 605, batch: 1, loss: 3.4232935905456543, modularity_loss: -0.6084868907928467, purity_loss: 3.957080364227295, regularization_loss: 0.07470013946294785 # 11:36:37 --- INFO: epoch: 606, batch: 1, loss: 3.422853469848633, modularity_loss: -0.6084998846054077, purity_loss: 3.9566521644592285, regularization_loss: 0.07470130175352097 # 11:36:37 --- INFO: epoch: 607, batch: 1, loss: 3.4224154949188232, modularity_loss: -0.6085119247436523, purity_loss: 3.9562253952026367, regularization_loss: 0.07470196485519409 # 11:36:37 --- INFO: epoch: 608, batch: 1, loss: 3.4219868183135986, modularity_loss: -0.6085251569747925, purity_loss: 3.9558098316192627, regularization_loss: 0.07470208406448364 # 11:36:38 --- INFO: epoch: 609, batch: 1, loss: 3.421563148498535, modularity_loss: -0.6085386276245117, purity_loss: 3.955399990081787, regularization_loss: 0.07470188289880753 # 11:36:38 --- INFO: epoch: 610, batch: 1, loss: 3.4211442470550537, modularity_loss: -0.6085527539253235, purity_loss: 3.9549951553344727, regularization_loss: 0.07470178604125977 # 11:36:38 --- INFO: epoch: 611, batch: 1, loss: 3.420729160308838, modularity_loss: -0.6085662245750427, purity_loss: 3.9545934200286865, regularization_loss: 0.07470201700925827 # 11:36:39 --- INFO: epoch: 612, batch: 1, loss: 3.420315742492676, modularity_loss: -0.6085794568061829, purity_loss: 3.954192638397217, regularization_loss: 0.07470262050628662 # 11:36:39 --- INFO: epoch: 613, batch: 1, loss: 3.4199066162109375, modularity_loss: -0.6085907220840454, purity_loss: 3.953793525695801, regularization_loss: 0.07470368593931198 # 11:36:39 --- INFO: epoch: 614, batch: 1, loss: 3.419501304626465, modularity_loss: -0.608601450920105, purity_loss: 3.953397750854492, regularization_loss: 0.07470505684614182 # 11:36:39 --- INFO: epoch: 615, batch: 1, loss: 3.419100284576416, modularity_loss: -0.6086120009422302, purity_loss: 3.9530060291290283, regularization_loss: 0.07470634579658508 # 11:36:40 --- INFO: epoch: 616, batch: 1, loss: 3.418704032897949, modularity_loss: -0.6086233854293823, purity_loss: 3.952620267868042, regularization_loss: 0.07470722496509552 # 11:36:40 --- INFO: epoch: 617, batch: 1, loss: 3.4183120727539062, modularity_loss: -0.6086347699165344, purity_loss: 3.952239513397217, regularization_loss: 0.07470735907554626 # 11:36:40 --- INFO: epoch: 618, batch: 1, loss: 3.417924404144287, modularity_loss: -0.6086475253105164, purity_loss: 3.9518649578094482, regularization_loss: 0.07470697909593582 # 11:36:41 --- INFO: epoch: 619, batch: 1, loss: 3.417537212371826, modularity_loss: -0.6086602807044983, purity_loss: 3.951491117477417, regularization_loss: 0.07470647245645523 # 11:36:41 --- INFO: epoch: 620, batch: 1, loss: 3.417156457901001, modularity_loss: -0.6086721420288086, purity_loss: 3.951122522354126, regularization_loss: 0.07470603287220001 # 11:36:41 --- INFO: epoch: 621, batch: 1, loss: 3.4167776107788086, modularity_loss: -0.6086831092834473, purity_loss: 3.9507548809051514, regularization_loss: 0.0747058317065239 # 11:36:42 --- INFO: epoch: 622, batch: 1, loss: 3.416400909423828, modularity_loss: -0.6086924076080322, purity_loss: 3.9503872394561768, regularization_loss: 0.07470603287220001 # 11:36:42 --- INFO: epoch: 623, batch: 1, loss: 3.4160304069519043, modularity_loss: -0.6086999177932739, purity_loss: 3.950023651123047, regularization_loss: 0.07470670342445374 # 11:36:42 --- INFO: epoch: 624, batch: 1, loss: 3.4156582355499268, modularity_loss: -0.6087060570716858, purity_loss: 3.9496567249298096, regularization_loss: 0.07470755279064178 # 11:36:42 --- INFO: epoch: 625, batch: 1, loss: 3.415294885635376, modularity_loss: -0.6087099313735962, purity_loss: 3.949296474456787, regularization_loss: 0.07470835000276566 # 11:36:43 --- INFO: epoch: 626, batch: 1, loss: 3.4149301052093506, modularity_loss: -0.6087135672569275, purity_loss: 3.94893479347229, regularization_loss: 0.07470890879631042 # 11:36:43 --- INFO: epoch: 627, batch: 1, loss: 3.4145689010620117, modularity_loss: -0.6087162494659424, purity_loss: 3.948575973510742, regularization_loss: 0.07470915466547012 # 11:36:43 --- INFO: epoch: 628, batch: 1, loss: 3.414212465286255, modularity_loss: -0.608718752861023, purity_loss: 3.9482221603393555, regularization_loss: 0.07470907270908356 # 11:36:44 --- INFO: epoch: 629, batch: 1, loss: 3.4138550758361816, modularity_loss: -0.6087208390235901, purity_loss: 3.9478671550750732, regularization_loss: 0.07470884919166565 # 11:36:44 --- INFO: epoch: 630, batch: 1, loss: 3.413503646850586, modularity_loss: -0.6087228059768677, purity_loss: 3.9475176334381104, regularization_loss: 0.07470874488353729 # 11:36:44 --- INFO: epoch: 631, batch: 1, loss: 3.4131507873535156, modularity_loss: -0.6087247133255005, purity_loss: 3.947166681289673, regularization_loss: 0.07470893114805222 # 11:36:45 --- INFO: epoch: 632, batch: 1, loss: 3.4128026962280273, modularity_loss: -0.6087263822555542, purity_loss: 3.94681978225708, regularization_loss: 0.07470938563346863 # 11:36:45 --- INFO: epoch: 633, batch: 1, loss: 3.412454605102539, modularity_loss: -0.6087270975112915, purity_loss: 3.946471691131592, regularization_loss: 0.07470990717411041 # 11:36:45 --- INFO: epoch: 634, batch: 1, loss: 3.4121100902557373, modularity_loss: -0.6087270975112915, purity_loss: 3.946126699447632, regularization_loss: 0.0747104212641716 # 11:36:46 --- INFO: epoch: 635, batch: 1, loss: 3.411769151687622, modularity_loss: -0.6087260246276855, purity_loss: 3.945784330368042, regularization_loss: 0.0747108981013298 # 11:36:46 --- INFO: epoch: 636, batch: 1, loss: 3.411428928375244, modularity_loss: -0.6087247133255005, purity_loss: 3.9454421997070312, regularization_loss: 0.07471131533384323 # 11:36:46 --- INFO: epoch: 637, batch: 1, loss: 3.411085367202759, modularity_loss: -0.6087231636047363, purity_loss: 3.945096969604492, regularization_loss: 0.07471158355474472 # 11:36:46 --- INFO: epoch: 638, batch: 1, loss: 3.410750389099121, modularity_loss: -0.6087213754653931, purity_loss: 3.9447600841522217, regularization_loss: 0.0747116282582283 # 11:36:47 --- INFO: epoch: 639, batch: 1, loss: 3.4104115962982178, modularity_loss: -0.6087194681167603, purity_loss: 3.9444196224212646, regularization_loss: 0.07471141964197159 # 11:36:47 --- INFO: epoch: 640, batch: 1, loss: 3.4100804328918457, modularity_loss: -0.6087167859077454, purity_loss: 3.9440858364105225, regularization_loss: 0.07471125572919846 # 11:36:47 --- INFO: epoch: 641, batch: 1, loss: 3.4097447395324707, modularity_loss: -0.6087149381637573, purity_loss: 3.9437484741210938, regularization_loss: 0.07471132278442383 # 11:36:48 --- INFO: epoch: 642, batch: 1, loss: 3.4094150066375732, modularity_loss: -0.6087134480476379, purity_loss: 3.9434168338775635, regularization_loss: 0.07471165060997009 # 11:36:48 --- INFO: epoch: 643, batch: 1, loss: 3.4090845584869385, modularity_loss: -0.6087138056755066, purity_loss: 3.9430863857269287, regularization_loss: 0.07471201568841934 # 11:36:48 --- INFO: epoch: 644, batch: 1, loss: 3.4087586402893066, modularity_loss: -0.6087148785591125, purity_loss: 3.942761182785034, regularization_loss: 0.07471229135990143 # 11:36:49 --- INFO: epoch: 645, batch: 1, loss: 3.408432960510254, modularity_loss: -0.6087170839309692, purity_loss: 3.9424374103546143, regularization_loss: 0.07471249252557755 # 11:36:49 --- INFO: epoch: 646, batch: 1, loss: 3.408106803894043, modularity_loss: -0.6087185144424438, purity_loss: 3.942112684249878, regularization_loss: 0.07471267879009247 # 11:36:49 --- INFO: epoch: 647, batch: 1, loss: 3.4077844619750977, modularity_loss: -0.6087194681167603, purity_loss: 3.94179105758667, regularization_loss: 0.07471282035112381 # 11:36:49 --- INFO: epoch: 648, batch: 1, loss: 3.4074599742889404, modularity_loss: -0.6087188720703125, purity_loss: 3.9414658546447754, regularization_loss: 0.07471293956041336 # 11:36:50 --- INFO: epoch: 649, batch: 1, loss: 3.4071412086486816, modularity_loss: -0.6087182760238647, purity_loss: 3.9411466121673584, regularization_loss: 0.07471296936273575 # 11:36:50 --- INFO: epoch: 650, batch: 1, loss: 3.4068222045898438, modularity_loss: -0.6087173223495483, purity_loss: 3.940826654434204, regularization_loss: 0.07471298426389694 # 11:36:50 --- INFO: epoch: 651, batch: 1, loss: 3.406503677368164, modularity_loss: -0.608717679977417, purity_loss: 3.9405081272125244, regularization_loss: 0.07471313327550888 # 11:36:51 --- INFO: epoch: 652, batch: 1, loss: 3.406187057495117, modularity_loss: -0.6087191700935364, purity_loss: 3.940192699432373, regularization_loss: 0.07471346110105515 # 11:36:51 --- INFO: epoch: 653, batch: 1, loss: 3.405871629714966, modularity_loss: -0.608721137046814, purity_loss: 3.9398789405822754, regularization_loss: 0.07471375167369843 # 11:36:51 --- INFO: epoch: 654, batch: 1, loss: 3.405555248260498, modularity_loss: -0.6087243556976318, purity_loss: 3.939565658569336, regularization_loss: 0.07471399009227753 # 11:36:51 --- INFO: epoch: 655, batch: 1, loss: 3.4052374362945557, modularity_loss: -0.6087278127670288, purity_loss: 3.939251184463501, regularization_loss: 0.07471403479576111 # 11:36:52 --- INFO: epoch: 656, batch: 1, loss: 3.404923915863037, modularity_loss: -0.6087294220924377, purity_loss: 3.938939332962036, regularization_loss: 0.07471392303705215 # 11:36:52 --- INFO: epoch: 657, batch: 1, loss: 3.4046080112457275, modularity_loss: -0.6087305545806885, purity_loss: 3.938624620437622, regularization_loss: 0.07471388578414917 # 11:36:52 --- INFO: epoch: 658, batch: 1, loss: 3.404294013977051, modularity_loss: -0.6087301969528198, purity_loss: 3.938310146331787, regularization_loss: 0.07471399754285812 # 11:36:53 --- INFO: epoch: 659, batch: 1, loss: 3.4039816856384277, modularity_loss: -0.6087299585342407, purity_loss: 3.937997579574585, regularization_loss: 0.07471414655447006 # 11:36:53 --- INFO: epoch: 660, batch: 1, loss: 3.4036686420440674, modularity_loss: -0.6087304353713989, purity_loss: 3.937685012817383, regularization_loss: 0.07471413165330887 # 11:36:53 --- INFO: epoch: 661, batch: 1, loss: 3.4033560752868652, modularity_loss: -0.6087322235107422, purity_loss: 3.9373741149902344, regularization_loss: 0.0747142806649208 # 11:36:54 --- INFO: epoch: 662, batch: 1, loss: 3.4030444622039795, modularity_loss: -0.6087340116500854, purity_loss: 3.937063694000244, regularization_loss: 0.07471473515033722 # 11:36:54 --- INFO: epoch: 663, batch: 1, loss: 3.4027369022369385, modularity_loss: -0.6087347269058228, purity_loss: 3.9367563724517822, regularization_loss: 0.07471530884504318 # 11:36:54 --- INFO: epoch: 664, batch: 1, loss: 3.4024248123168945, modularity_loss: -0.6087340712547302, purity_loss: 3.9364430904388428, regularization_loss: 0.07471586018800735 # 11:36:55 --- INFO: epoch: 665, batch: 1, loss: 3.402114152908325, modularity_loss: -0.6087313890457153, purity_loss: 3.936129331588745, regularization_loss: 0.07471615821123123 # 11:36:55 --- INFO: epoch: 666, batch: 1, loss: 3.4018073081970215, modularity_loss: -0.6087266206741333, purity_loss: 3.9358177185058594, regularization_loss: 0.07471610605716705 # 11:36:55 --- INFO: epoch: 667, batch: 1, loss: 3.401498556137085, modularity_loss: -0.6087207794189453, purity_loss: 3.9355034828186035, regularization_loss: 0.07471592724323273 # 11:36:56 --- INFO: epoch: 668, batch: 1, loss: 3.4011902809143066, modularity_loss: -0.6087154150009155, purity_loss: 3.935189962387085, regularization_loss: 0.0747157633304596 # 11:36:56 --- INFO: epoch: 669, batch: 1, loss: 3.4008853435516357, modularity_loss: -0.6087104082107544, purity_loss: 3.934880256652832, regularization_loss: 0.07471555471420288 # 11:36:56 --- INFO: epoch: 670, batch: 1, loss: 3.4005777835845947, modularity_loss: -0.6087072491645813, purity_loss: 3.9345695972442627, regularization_loss: 0.07471540570259094 # 11:36:56 --- INFO: epoch: 671, batch: 1, loss: 3.4002740383148193, modularity_loss: -0.6087055206298828, purity_loss: 3.9342641830444336, regularization_loss: 0.07471542060375214 # 11:36:57 --- INFO: epoch: 672, batch: 1, loss: 3.399968385696411, modularity_loss: -0.6087043285369873, purity_loss: 3.933957099914551, regularization_loss: 0.07471569627523422 # 11:36:57 --- INFO: epoch: 673, batch: 1, loss: 3.399667501449585, modularity_loss: -0.6087037324905396, purity_loss: 3.933655023574829, regularization_loss: 0.07471621036529541 # 11:36:57 --- INFO: epoch: 674, batch: 1, loss: 3.3993635177612305, modularity_loss: -0.6087017059326172, purity_loss: 3.9333484172821045, regularization_loss: 0.07471665740013123 # 11:36:58 --- INFO: epoch: 675, batch: 1, loss: 3.399059295654297, modularity_loss: -0.608698308467865, purity_loss: 3.9330408573150635, regularization_loss: 0.07471680641174316 # 11:36:58 --- INFO: epoch: 676, batch: 1, loss: 3.3987560272216797, modularity_loss: -0.6086939573287964, purity_loss: 3.9327330589294434, regularization_loss: 0.07471685111522675 # 11:36:58 --- INFO: epoch: 677, batch: 1, loss: 3.398449420928955, modularity_loss: -0.6086899042129517, purity_loss: 3.932422399520874, regularization_loss: 0.07471691817045212 # 11:36:59 --- INFO: epoch: 678, batch: 1, loss: 3.3981475830078125, modularity_loss: -0.6086863875389099, purity_loss: 3.932116985321045, regularization_loss: 0.07471710443496704 # 11:36:59 --- INFO: epoch: 679, batch: 1, loss: 3.397845506668091, modularity_loss: -0.6086824536323547, purity_loss: 3.9318106174468994, regularization_loss: 0.07471736520528793 # 11:36:59 --- INFO: epoch: 680, batch: 1, loss: 3.3975448608398438, modularity_loss: -0.6086785197257996, purity_loss: 3.9315059185028076, regularization_loss: 0.07471759617328644 # 11:37:00 --- INFO: epoch: 681, batch: 1, loss: 3.3972411155700684, modularity_loss: -0.6086742281913757, purity_loss: 3.931197166442871, regularization_loss: 0.07471803575754166 # 11:37:00 --- INFO: epoch: 682, batch: 1, loss: 3.396941661834717, modularity_loss: -0.6086704134941101, purity_loss: 3.9308934211730957, regularization_loss: 0.0747186616063118 # 11:37:00 --- INFO: epoch: 683, batch: 1, loss: 3.3966407775878906, modularity_loss: -0.6086674928665161, purity_loss: 3.930589199066162, regularization_loss: 0.07471922039985657 # 11:37:01 --- INFO: epoch: 684, batch: 1, loss: 3.396338939666748, modularity_loss: -0.6086655259132385, purity_loss: 3.9302852153778076, regularization_loss: 0.0747193694114685 # 11:37:01 --- INFO: epoch: 685, batch: 1, loss: 3.3960366249084473, modularity_loss: -0.6086630821228027, purity_loss: 3.929980516433716, regularization_loss: 0.07471917569637299 # 11:37:01 --- INFO: epoch: 686, batch: 1, loss: 3.395739793777466, modularity_loss: -0.6086611747741699, purity_loss: 3.9296820163726807, regularization_loss: 0.0747189074754715 # 11:37:01 --- INFO: epoch: 687, batch: 1, loss: 3.3954405784606934, modularity_loss: -0.6086595058441162, purity_loss: 3.9293811321258545, regularization_loss: 0.0747188851237297 # 11:37:02 --- INFO: epoch: 688, batch: 1, loss: 3.395142078399658, modularity_loss: -0.608657717704773, purity_loss: 3.9290807247161865, regularization_loss: 0.07471900433301926 # 11:37:02 --- INFO: epoch: 689, batch: 1, loss: 3.394845724105835, modularity_loss: -0.6086556315422058, purity_loss: 3.9287824630737305, regularization_loss: 0.07471895962953568 # 11:37:02 --- INFO: epoch: 690, batch: 1, loss: 3.3945441246032715, modularity_loss: -0.60865318775177, purity_loss: 3.928478479385376, regularization_loss: 0.07471881061792374 # 11:37:03 --- INFO: epoch: 691, batch: 1, loss: 3.3942506313323975, modularity_loss: -0.6086512804031372, purity_loss: 3.928183078765869, regularization_loss: 0.07471885532140732 # 11:37:03 --- INFO: epoch: 692, batch: 1, loss: 3.393951654434204, modularity_loss: -0.608650803565979, purity_loss: 3.9278831481933594, regularization_loss: 0.07471923530101776 # 11:37:03 --- INFO: epoch: 693, batch: 1, loss: 3.3936567306518555, modularity_loss: -0.6086493134498596, purity_loss: 3.927586555480957, regularization_loss: 0.07471950352191925 # 11:37:04 --- INFO: epoch: 694, batch: 1, loss: 3.3933615684509277, modularity_loss: -0.608647882938385, purity_loss: 3.9272899627685547, regularization_loss: 0.07471954822540283 # 11:37:04 --- INFO: epoch: 695, batch: 1, loss: 3.3930654525756836, modularity_loss: -0.6086455583572388, purity_loss: 3.9269917011260986, regularization_loss: 0.0747193992137909 # 11:37:04 --- INFO: epoch: 696, batch: 1, loss: 3.392773151397705, modularity_loss: -0.6086423397064209, purity_loss: 3.926696300506592, regularization_loss: 0.07471930980682373 # 11:37:05 --- INFO: epoch: 697, batch: 1, loss: 3.392482280731201, modularity_loss: -0.6086400747299194, purity_loss: 3.9264028072357178, regularization_loss: 0.07471951842308044 # 11:37:05 --- INFO: epoch: 698, batch: 1, loss: 3.3921918869018555, modularity_loss: -0.6086380481719971, purity_loss: 3.92611026763916, regularization_loss: 0.07471972703933716 # 11:37:05 --- INFO: epoch: 699, batch: 1, loss: 3.391902446746826, modularity_loss: -0.6086361408233643, purity_loss: 3.925818681716919, regularization_loss: 0.07471977919340134 # 11:37:05 --- INFO: epoch: 700, batch: 1, loss: 3.3916144371032715, modularity_loss: -0.6086333394050598, purity_loss: 3.925528049468994, regularization_loss: 0.0747196301817894 # 11:37:06 --- INFO: epoch: 701, batch: 1, loss: 3.391327381134033, modularity_loss: -0.608630895614624, purity_loss: 3.925238609313965, regularization_loss: 0.07471958547830582 # 11:37:06 --- INFO: epoch: 702, batch: 1, loss: 3.3910417556762695, modularity_loss: -0.6086287498474121, purity_loss: 3.9249510765075684, regularization_loss: 0.07471953332424164 # 11:37:06 --- INFO: epoch: 703, batch: 1, loss: 3.3907575607299805, modularity_loss: -0.6086267828941345, purity_loss: 3.9246649742126465, regularization_loss: 0.07471925765275955 # 11:37:07 --- INFO: epoch: 704, batch: 1, loss: 3.390472888946533, modularity_loss: -0.6086251735687256, purity_loss: 3.9243791103363037, regularization_loss: 0.07471885532140732 # 11:37:07 --- INFO: epoch: 705, batch: 1, loss: 3.390190362930298, modularity_loss: -0.6086251139640808, purity_loss: 3.9240968227386475, regularization_loss: 0.07471857964992523 # 11:37:07 --- INFO: epoch: 706, batch: 1, loss: 3.3899097442626953, modularity_loss: -0.6086257100105286, purity_loss: 3.9238169193267822, regularization_loss: 0.0747186467051506 # 11:37:08 --- INFO: epoch: 707, batch: 1, loss: 3.3896307945251465, modularity_loss: -0.6086264848709106, purity_loss: 3.9235382080078125, regularization_loss: 0.07471904158592224 # 11:37:08 --- INFO: epoch: 708, batch: 1, loss: 3.389353036880493, modularity_loss: -0.6086269021034241, purity_loss: 3.923260450363159, regularization_loss: 0.07471946626901627 # 11:37:08 --- INFO: epoch: 709, batch: 1, loss: 3.38907527923584, modularity_loss: -0.6086262464523315, purity_loss: 3.9229817390441895, regularization_loss: 0.07471976429224014 # 11:37:09 --- INFO: epoch: 710, batch: 1, loss: 3.388794422149658, modularity_loss: -0.6086249947547913, purity_loss: 3.922699451446533, regularization_loss: 0.07472008466720581 # 11:37:09 --- INFO: epoch: 711, batch: 1, loss: 3.3885152339935303, modularity_loss: -0.6086239814758301, purity_loss: 3.9224188327789307, regularization_loss: 0.0747203528881073 # 11:37:09 --- INFO: epoch: 712, batch: 1, loss: 3.3882429599761963, modularity_loss: -0.6086227893829346, purity_loss: 3.922145366668701, regularization_loss: 0.07472046464681625 # 11:37:09 --- INFO: epoch: 713, batch: 1, loss: 3.3879714012145996, modularity_loss: -0.6086219549179077, purity_loss: 3.921872854232788, regularization_loss: 0.07472041994333267 # 11:37:10 --- INFO: epoch: 714, batch: 1, loss: 3.3877007961273193, modularity_loss: -0.6086221933364868, purity_loss: 3.921602725982666, regularization_loss: 0.07472028583288193 # 11:37:10 --- INFO: epoch: 715, batch: 1, loss: 3.387432336807251, modularity_loss: -0.6086224317550659, purity_loss: 3.9213345050811768, regularization_loss: 0.07472030073404312 # 11:37:10 --- INFO: epoch: 716, batch: 1, loss: 3.387162446975708, modularity_loss: -0.6086228489875793, purity_loss: 3.921064853668213, regularization_loss: 0.07472050189971924 # 11:37:11 --- INFO: epoch: 717, batch: 1, loss: 3.3868978023529053, modularity_loss: -0.6086224317550659, purity_loss: 3.920799493789673, regularization_loss: 0.07472069561481476 # 11:37:11 --- INFO: epoch: 718, batch: 1, loss: 3.38663387298584, modularity_loss: -0.6086207628250122, purity_loss: 3.9205338954925537, regularization_loss: 0.07472078502178192 # 11:37:11 --- INFO: epoch: 719, batch: 1, loss: 3.3863699436187744, modularity_loss: -0.608618974685669, purity_loss: 3.9202682971954346, regularization_loss: 0.0747205838561058 # 11:37:12 --- INFO: epoch: 720, batch: 1, loss: 3.386107921600342, modularity_loss: -0.608617901802063, purity_loss: 3.9200053215026855, regularization_loss: 0.07472046464681625 # 11:37:12 --- INFO: epoch: 721, batch: 1, loss: 3.3858487606048584, modularity_loss: -0.6086181402206421, purity_loss: 3.9197463989257812, regularization_loss: 0.07472053170204163 # 11:37:12 --- INFO: epoch: 722, batch: 1, loss: 3.385589599609375, modularity_loss: -0.6086193323135376, purity_loss: 3.9194881916046143, regularization_loss: 0.07472065091133118 # 11:37:12 --- INFO: epoch: 723, batch: 1, loss: 3.385335922241211, modularity_loss: -0.6086205244064331, purity_loss: 3.9192357063293457, regularization_loss: 0.07472066581249237 # 11:37:13 --- INFO: epoch: 724, batch: 1, loss: 3.3850817680358887, modularity_loss: -0.6086221933364868, purity_loss: 3.918982982635498, regularization_loss: 0.0747208446264267 # 11:37:13 --- INFO: epoch: 725, batch: 1, loss: 3.384828805923462, modularity_loss: -0.6086236834526062, purity_loss: 3.918731212615967, regularization_loss: 0.0747213140130043 # 11:37:13 --- INFO: epoch: 726, batch: 1, loss: 3.3845765590667725, modularity_loss: -0.6086239814758301, purity_loss: 3.9184789657592773, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 727, batch: 1, loss: 3.3843271732330322, modularity_loss: -0.6086238622665405, purity_loss: 3.918229341506958, regularization_loss: 0.07472165673971176 # 11:37:14 --- INFO: epoch: 728, batch: 1, loss: 3.3840792179107666, modularity_loss: -0.6086242198944092, purity_loss: 3.9179821014404297, regularization_loss: 0.07472139596939087 # 11:37:14 --- INFO: epoch: 729, batch: 1, loss: 3.3838346004486084, modularity_loss: -0.6086263656616211, purity_loss: 3.9177398681640625, regularization_loss: 0.0747210681438446 # 11:37:15 --- INFO: epoch: 730, batch: 1, loss: 3.3835926055908203, modularity_loss: -0.6086296439170837, purity_loss: 3.917501449584961, regularization_loss: 0.0747208297252655 # 11:37:15 --- INFO: epoch: 731, batch: 1, loss: 3.3833510875701904, modularity_loss: -0.608633279800415, purity_loss: 3.9172637462615967, regularization_loss: 0.07472065836191177 # 11:37:15 --- INFO: epoch: 732, batch: 1, loss: 3.3831124305725098, modularity_loss: -0.6086359024047852, purity_loss: 3.917027711868286, regularization_loss: 0.07472063601016998 # 11:37:15 --- INFO: epoch: 733, batch: 1, loss: 3.382878065109253, modularity_loss: -0.6086368560791016, purity_loss: 3.9167940616607666, regularization_loss: 0.07472086697816849 # 11:37:16 --- INFO: epoch: 734, batch: 1, loss: 3.382638931274414, modularity_loss: -0.6086379289627075, purity_loss: 3.916555643081665, regularization_loss: 0.07472129166126251 # 11:37:16 --- INFO: epoch: 735, batch: 1, loss: 3.382408618927002, modularity_loss: -0.6086374521255493, purity_loss: 3.9163243770599365, regularization_loss: 0.07472161948680878 # 11:37:16 --- INFO: epoch: 736, batch: 1, loss: 3.3821797370910645, modularity_loss: -0.608637273311615, purity_loss: 3.916095495223999, regularization_loss: 0.07472164183855057 # 11:37:17 --- INFO: epoch: 737, batch: 1, loss: 3.381951332092285, modularity_loss: -0.6086390614509583, purity_loss: 3.9158689975738525, regularization_loss: 0.07472144067287445 # 11:37:17 --- INFO: epoch: 738, batch: 1, loss: 3.381723403930664, modularity_loss: -0.6086426973342896, purity_loss: 3.915644884109497, regularization_loss: 0.07472117990255356 # 11:37:17 --- INFO: epoch: 739, batch: 1, loss: 3.3814990520477295, modularity_loss: -0.6086472272872925, purity_loss: 3.9154253005981445, regularization_loss: 0.07472096383571625 # 11:37:18 --- INFO: epoch: 740, batch: 1, loss: 3.381277561187744, modularity_loss: -0.6086505651473999, purity_loss: 3.9152073860168457, regularization_loss: 0.07472078502178192 # 11:37:18 --- INFO: epoch: 741, batch: 1, loss: 3.3810572624206543, modularity_loss: -0.6086528301239014, purity_loss: 3.914989471435547, regularization_loss: 0.07472071796655655 # 11:37:18 --- INFO: epoch: 742, batch: 1, loss: 3.38084077835083, modularity_loss: -0.6086551547050476, purity_loss: 3.9147748947143555, regularization_loss: 0.07472089678049088 # 11:37:18 --- INFO: epoch: 743, batch: 1, loss: 3.3806259632110596, modularity_loss: -0.6086583137512207, purity_loss: 3.914562940597534, regularization_loss: 0.07472134381532669 # 11:37:19 --- INFO: epoch: 744, batch: 1, loss: 3.3804121017456055, modularity_loss: -0.6086623668670654, purity_loss: 3.9143528938293457, regularization_loss: 0.07472160458564758 # 11:37:19 --- INFO: epoch: 745, batch: 1, loss: 3.380201816558838, modularity_loss: -0.6086667776107788, purity_loss: 3.914146900177002, regularization_loss: 0.07472165673971176 # 11:37:19 --- INFO: epoch: 746, batch: 1, loss: 3.379990339279175, modularity_loss: -0.6086707711219788, purity_loss: 3.9139394760131836, regularization_loss: 0.07472170144319534 # 11:37:20 --- INFO: epoch: 747, batch: 1, loss: 3.3797826766967773, modularity_loss: -0.6086737513542175, purity_loss: 3.9137344360351562, regularization_loss: 0.07472185045480728 # 11:37:20 --- INFO: epoch: 748, batch: 1, loss: 3.3795764446258545, modularity_loss: -0.6086753606796265, purity_loss: 3.913529872894287, regularization_loss: 0.07472192496061325 # 11:37:20 --- INFO: epoch: 749, batch: 1, loss: 3.3793721199035645, modularity_loss: -0.6086761355400085, purity_loss: 3.9133262634277344, regularization_loss: 0.07472185045480728 # 11:37:21 --- INFO: epoch: 750, batch: 1, loss: 3.3791720867156982, modularity_loss: -0.6086779832839966, purity_loss: 3.91312837600708, regularization_loss: 0.07472176104784012 # 11:37:21 --- INFO: epoch: 751, batch: 1, loss: 3.3789730072021484, modularity_loss: -0.6086809635162354, purity_loss: 3.9129321575164795, regularization_loss: 0.07472184300422668 # 11:37:21 --- INFO: epoch: 752, batch: 1, loss: 3.3787708282470703, modularity_loss: -0.6086852550506592, purity_loss: 3.912734270095825, regularization_loss: 0.07472191751003265 # 11:37:21 --- INFO: epoch: 753, batch: 1, loss: 3.378572702407837, modularity_loss: -0.6086894273757935, purity_loss: 3.9125401973724365, regularization_loss: 0.07472185045480728 # 11:37:22 --- INFO: epoch: 754, batch: 1, loss: 3.3783771991729736, modularity_loss: -0.6086921691894531, purity_loss: 3.9123475551605225, regularization_loss: 0.07472173869609833 # 11:37:22 --- INFO: epoch: 755, batch: 1, loss: 3.3781824111938477, modularity_loss: -0.6086947917938232, purity_loss: 3.9121556282043457, regularization_loss: 0.074721559882164 # 11:37:22 --- INFO: epoch: 756, batch: 1, loss: 3.3779866695404053, modularity_loss: -0.6086962819099426, purity_loss: 3.911961555480957, regularization_loss: 0.07472142577171326 # 11:37:23 --- INFO: epoch: 757, batch: 1, loss: 3.3777947425842285, modularity_loss: -0.6086974143981934, purity_loss: 3.911770820617676, regularization_loss: 0.07472137361764908 # 11:37:23 --- INFO: epoch: 758, batch: 1, loss: 3.377608060836792, modularity_loss: -0.6086984872817993, purity_loss: 3.9115853309631348, regularization_loss: 0.07472129166126251 # 11:37:23 --- INFO: epoch: 759, batch: 1, loss: 3.3774170875549316, modularity_loss: -0.6086995601654053, purity_loss: 3.911395311355591, regularization_loss: 0.07472124695777893 # 11:37:24 --- INFO: epoch: 760, batch: 1, loss: 3.377230167388916, modularity_loss: -0.6086998581886292, purity_loss: 3.9112091064453125, regularization_loss: 0.07472096383571625 # 11:37:24 --- INFO: epoch: 761, batch: 1, loss: 3.377042293548584, modularity_loss: -0.608700156211853, purity_loss: 3.9110217094421387, regularization_loss: 0.07472078502178192 # 11:37:24 --- INFO: epoch: 762, batch: 1, loss: 3.3768577575683594, modularity_loss: -0.6087008714675903, purity_loss: 3.9108376502990723, regularization_loss: 0.0747208446264267 # 11:37:24 --- INFO: epoch: 763, batch: 1, loss: 3.376673698425293, modularity_loss: -0.6087024211883545, purity_loss: 3.9106552600860596, regularization_loss: 0.07472086697816849 # 11:37:25 --- INFO: epoch: 764, batch: 1, loss: 3.376494884490967, modularity_loss: -0.6087031364440918, purity_loss: 3.91047739982605, regularization_loss: 0.07472074776887894 # 11:37:25 --- INFO: epoch: 765, batch: 1, loss: 3.3763132095336914, modularity_loss: -0.6087043285369873, purity_loss: 3.910296678543091, regularization_loss: 0.07472091913223267 # 11:37:25 --- INFO: epoch: 766, batch: 1, loss: 3.376133680343628, modularity_loss: -0.6087051630020142, purity_loss: 3.9101176261901855, regularization_loss: 0.07472129911184311 # 11:37:26 --- INFO: epoch: 767, batch: 1, loss: 3.375957489013672, modularity_loss: -0.6087040305137634, purity_loss: 3.909940004348755, regularization_loss: 0.07472151517868042 # 11:37:26 --- INFO: epoch: 768, batch: 1, loss: 3.3757801055908203, modularity_loss: -0.6087014675140381, purity_loss: 3.9097602367401123, regularization_loss: 0.07472142577171326 # 11:37:26 --- INFO: epoch: 769, batch: 1, loss: 3.375605821609497, modularity_loss: -0.6086981296539307, purity_loss: 3.9095826148986816, regularization_loss: 0.07472129166126251 # 11:37:27 --- INFO: epoch: 770, batch: 1, loss: 3.3754348754882812, modularity_loss: -0.6086962223052979, purity_loss: 3.909409761428833, regularization_loss: 0.07472126185894012 # 11:37:27 --- INFO: epoch: 771, batch: 1, loss: 3.3752622604370117, modularity_loss: -0.6086959838867188, purity_loss: 3.9092373847961426, regularization_loss: 0.07472093403339386 # 11:37:27 --- INFO: epoch: 772, batch: 1, loss: 3.375091552734375, modularity_loss: -0.6086970567703247, purity_loss: 3.9090678691864014, regularization_loss: 0.07472062855958939 # 11:37:27 --- INFO: epoch: 773, batch: 1, loss: 3.3749241828918457, modularity_loss: -0.608698844909668, purity_loss: 3.908902406692505, regularization_loss: 0.07472056895494461 # 11:37:28 --- INFO: epoch: 774, batch: 1, loss: 3.374755859375, modularity_loss: -0.6087007522583008, purity_loss: 3.908735990524292, regularization_loss: 0.07472071796655655 # 11:37:28 --- INFO: epoch: 775, batch: 1, loss: 3.3745923042297363, modularity_loss: -0.6087013483047485, purity_loss: 3.9085726737976074, regularization_loss: 0.07472091913223267 # 11:37:28 --- INFO: epoch: 776, batch: 1, loss: 3.3744232654571533, modularity_loss: -0.6087007522583008, purity_loss: 3.908402919769287, regularization_loss: 0.07472109794616699 # 11:37:29 --- INFO: epoch: 777, batch: 1, loss: 3.374260902404785, modularity_loss: -0.608699381351471, purity_loss: 3.9082393646240234, regularization_loss: 0.07472093403339386 # 11:37:29 --- INFO: epoch: 778, batch: 1, loss: 3.3740997314453125, modularity_loss: -0.6086994409561157, purity_loss: 3.90807843208313, regularization_loss: 0.0747208297252655 # 11:37:29 --- INFO: epoch: 779, batch: 1, loss: 3.3739380836486816, modularity_loss: -0.608701229095459, purity_loss: 3.9079184532165527, regularization_loss: 0.07472071796655655 # 11:37:29 --- INFO: epoch: 780, batch: 1, loss: 3.373779773712158, modularity_loss: -0.6087039709091187, purity_loss: 3.9077632427215576, regularization_loss: 0.07472056895494461 # 11:37:30 --- INFO: epoch: 781, batch: 1, loss: 3.373619556427002, modularity_loss: -0.608706533908844, purity_loss: 3.9076056480407715, regularization_loss: 0.07472051680088043 # 11:37:30 --- INFO: epoch: 782, batch: 1, loss: 3.3734633922576904, modularity_loss: -0.6087087392807007, purity_loss: 3.9074513912200928, regularization_loss: 0.07472073286771774 # 11:37:30 --- INFO: epoch: 783, batch: 1, loss: 3.373309373855591, modularity_loss: -0.6087095737457275, purity_loss: 3.9072978496551514, regularization_loss: 0.07472101598978043 # 11:37:31 --- INFO: epoch: 784, batch: 1, loss: 3.373154401779175, modularity_loss: -0.6087086796760559, purity_loss: 3.907142162322998, regularization_loss: 0.07472088187932968 # 11:37:31 --- INFO: epoch: 785, batch: 1, loss: 3.372997999191284, modularity_loss: -0.6087080240249634, purity_loss: 3.906985282897949, regularization_loss: 0.07472073286771774 # 11:37:31 --- INFO: epoch: 786, batch: 1, loss: 3.3728485107421875, modularity_loss: -0.6087089776992798, purity_loss: 3.906836748123169, regularization_loss: 0.0747205913066864 # 11:37:31 --- INFO: epoch: 787, batch: 1, loss: 3.37269926071167, modularity_loss: -0.6087101697921753, purity_loss: 3.906688928604126, regularization_loss: 0.07472040504217148 # 11:37:32 --- INFO: epoch: 788, batch: 1, loss: 3.3725497722625732, modularity_loss: -0.6087112426757812, purity_loss: 3.906540632247925, regularization_loss: 0.0747203379869461 # 11:37:32 --- INFO: epoch: 789, batch: 1, loss: 3.372401237487793, modularity_loss: -0.6087126731872559, purity_loss: 3.906393527984619, regularization_loss: 0.07472032308578491 # 11:37:32 --- INFO: epoch: 790, batch: 1, loss: 3.372250556945801, modularity_loss: -0.6087139844894409, purity_loss: 3.9062440395355225, regularization_loss: 0.07472048699855804 # 11:37:33 --- INFO: epoch: 791, batch: 1, loss: 3.3721022605895996, modularity_loss: -0.6087154746055603, purity_loss: 3.906097173690796, regularization_loss: 0.07472061365842819 # 11:37:33 --- INFO: epoch: 792, batch: 1, loss: 3.3719561100006104, modularity_loss: -0.6087169647216797, purity_loss: 3.9059524536132812, regularization_loss: 0.07472064346075058 # 11:37:33 --- INFO: epoch: 793, batch: 1, loss: 3.3718109130859375, modularity_loss: -0.6087177991867065, purity_loss: 3.905808210372925, regularization_loss: 0.07472051680088043 # 11:37:34 --- INFO: epoch: 794, batch: 1, loss: 3.37166690826416, modularity_loss: -0.6087191104888916, purity_loss: 3.905665874481201, regularization_loss: 0.07472027093172073 # 11:37:34 --- INFO: epoch: 795, batch: 1, loss: 3.371525764465332, modularity_loss: -0.6087210774421692, purity_loss: 3.905526638031006, regularization_loss: 0.07472021877765656 # 11:37:34 --- INFO: epoch: 796, batch: 1, loss: 3.371382236480713, modularity_loss: -0.6087223291397095, purity_loss: 3.9053843021392822, regularization_loss: 0.07472039759159088 # 11:37:34 --- INFO: epoch: 797, batch: 1, loss: 3.371239423751831, modularity_loss: -0.6087222099304199, purity_loss: 3.9052412509918213, regularization_loss: 0.07472032308578491 # 11:37:35 --- INFO: epoch: 798, batch: 1, loss: 3.3710997104644775, modularity_loss: -0.6087217926979065, purity_loss: 3.9051010608673096, regularization_loss: 0.07472048699855804 # 11:37:35 --- INFO: epoch: 799, batch: 1, loss: 3.3709588050842285, modularity_loss: -0.6087220907211304, purity_loss: 3.9049599170684814, regularization_loss: 0.07472088187932968 # 11:37:35 --- INFO: epoch: 800, batch: 1, loss: 3.370821475982666, modularity_loss: -0.6087215542793274, purity_loss: 3.9048221111297607, regularization_loss: 0.07472094893455505 # 11:37:36 --- INFO: epoch: 801, batch: 1, loss: 3.370682716369629, modularity_loss: -0.6087210178375244, purity_loss: 3.9046831130981445, regularization_loss: 0.07472068071365356 # 11:37:36 --- INFO: epoch: 802, batch: 1, loss: 3.37054443359375, modularity_loss: -0.608720064163208, purity_loss: 3.9045441150665283, regularization_loss: 0.07472044974565506 # 11:37:36 --- INFO: epoch: 803, batch: 1, loss: 3.370408296585083, modularity_loss: -0.6087198257446289, purity_loss: 3.9044077396392822, regularization_loss: 0.07472046464681625 # 11:37:37 --- INFO: epoch: 804, batch: 1, loss: 3.3702731132507324, modularity_loss: -0.6087179780006409, purity_loss: 3.904270648956299, regularization_loss: 0.07472040504217148 # 11:37:37 --- INFO: epoch: 805, batch: 1, loss: 3.370136022567749, modularity_loss: -0.6087155938148499, purity_loss: 3.9041314125061035, regularization_loss: 0.07472021877765656 # 11:37:37 --- INFO: epoch: 806, batch: 1, loss: 3.370004415512085, modularity_loss: -0.6087132096290588, purity_loss: 3.9039974212646484, regularization_loss: 0.07472027093172073 # 11:37:37 --- INFO: epoch: 807, batch: 1, loss: 3.3698718547821045, modularity_loss: -0.6087116003036499, purity_loss: 3.903862953186035, regularization_loss: 0.07472051680088043 # 11:37:38 --- INFO: epoch: 808, batch: 1, loss: 3.3697381019592285, modularity_loss: -0.6087098717689514, purity_loss: 3.9037275314331055, regularization_loss: 0.0747205913066864 # 11:37:38 --- INFO: epoch: 809, batch: 1, loss: 3.3696084022521973, modularity_loss: -0.6087083220481873, purity_loss: 3.9035964012145996, regularization_loss: 0.07472043484449387 # 11:37:38 --- INFO: epoch: 810, batch: 1, loss: 3.3694772720336914, modularity_loss: -0.6087076663970947, purity_loss: 3.9034645557403564, regularization_loss: 0.0747203677892685 # 11:37:39 --- INFO: epoch: 811, batch: 1, loss: 3.3693482875823975, modularity_loss: -0.6087066531181335, purity_loss: 3.903334617614746, regularization_loss: 0.0747203528881073 # 11:37:39 --- INFO: epoch: 812, batch: 1, loss: 3.36921763420105, modularity_loss: -0.6087048053741455, purity_loss: 3.9032022953033447, regularization_loss: 0.07472015917301178 # 11:37:39 --- INFO: epoch: 813, batch: 1, loss: 3.3690872192382812, modularity_loss: -0.6087033152580261, purity_loss: 3.9030704498291016, regularization_loss: 0.07471994310617447 # 11:37:40 --- INFO: epoch: 814, batch: 1, loss: 3.36896014213562, modularity_loss: -0.608702540397644, purity_loss: 3.902942657470703, regularization_loss: 0.07471997290849686 # 11:37:40 --- INFO: epoch: 815, batch: 1, loss: 3.368832588195801, modularity_loss: -0.6087023019790649, purity_loss: 3.9028146266937256, regularization_loss: 0.07472013682126999 # 11:37:40 --- INFO: epoch: 816, batch: 1, loss: 3.3687071800231934, modularity_loss: -0.6087015867233276, purity_loss: 3.902688503265381, regularization_loss: 0.0747201219201088 # 11:37:40 --- INFO: epoch: 817, batch: 1, loss: 3.368579387664795, modularity_loss: -0.6087002754211426, purity_loss: 3.902559757232666, regularization_loss: 0.07471991330385208 # 11:37:41 --- INFO: epoch: 818, batch: 1, loss: 3.368454933166504, modularity_loss: -0.6086994409561157, purity_loss: 3.9024345874786377, regularization_loss: 0.07471971213817596 # 11:37:41 --- INFO: epoch: 819, batch: 1, loss: 3.368333339691162, modularity_loss: -0.6086984276771545, purity_loss: 3.9023122787475586, regularization_loss: 0.0747196301817894 # 11:37:41 --- INFO: epoch: 820, batch: 1, loss: 3.368211030960083, modularity_loss: -0.6086968183517456, purity_loss: 3.902188301086426, regularization_loss: 0.07471960037946701 # 11:37:42 --- INFO: epoch: 821, batch: 1, loss: 3.368088722229004, modularity_loss: -0.6086952686309814, purity_loss: 3.902064323425293, regularization_loss: 0.0747196227312088 # 11:37:42 --- INFO: epoch: 822, batch: 1, loss: 3.3679676055908203, modularity_loss: -0.6086933612823486, purity_loss: 3.9019412994384766, regularization_loss: 0.07471977919340134 # 11:37:42 --- INFO: epoch: 823, batch: 1, loss: 3.367844343185425, modularity_loss: -0.6086910963058472, purity_loss: 3.901815414428711, regularization_loss: 0.07472006976604462 # 11:37:43 --- INFO: epoch: 824, batch: 1, loss: 3.367724895477295, modularity_loss: -0.6086887121200562, purity_loss: 3.901693344116211, regularization_loss: 0.07472028583288193 # 11:37:43 --- INFO: epoch: 825, batch: 1, loss: 3.367605447769165, modularity_loss: -0.6086865663528442, purity_loss: 3.901571750640869, regularization_loss: 0.07472020387649536 # 11:37:43 --- INFO: epoch: 826, batch: 1, loss: 3.367486000061035, modularity_loss: -0.6086844801902771, purity_loss: 3.9014506340026855, regularization_loss: 0.07471991330385208 # 11:37:43 --- INFO: epoch: 827, batch: 1, loss: 3.3673691749572754, modularity_loss: -0.6086835861206055, purity_loss: 3.9013330936431885, regularization_loss: 0.0747196227312088 # 11:37:44 --- INFO: epoch: 828, batch: 1, loss: 3.3672499656677246, modularity_loss: -0.6086843013763428, purity_loss: 3.901214838027954, regularization_loss: 0.0747193992137909 # 11:37:44 --- INFO: epoch: 829, batch: 1, loss: 3.3671350479125977, modularity_loss: -0.6086848378181458, purity_loss: 3.9011006355285645, regularization_loss: 0.0747193917632103 # 11:37:44 --- INFO: epoch: 830, batch: 1, loss: 3.367018699645996, modularity_loss: -0.6086846590042114, purity_loss: 3.9009838104248047, regularization_loss: 0.0747196152806282 # 11:37:45 --- INFO: epoch: 831, batch: 1, loss: 3.3669023513793945, modularity_loss: -0.6086817979812622, purity_loss: 3.900864362716675, regularization_loss: 0.07471993565559387 # 11:37:45 --- INFO: epoch: 832, batch: 1, loss: 3.366786241531372, modularity_loss: -0.6086785793304443, purity_loss: 3.900744676589966, regularization_loss: 0.07472006976604462 # 11:37:45 --- INFO: epoch: 833, batch: 1, loss: 3.366670846939087, modularity_loss: -0.6086764335632324, purity_loss: 3.900627374649048, regularization_loss: 0.07471989095211029 # 11:37:46 --- INFO: epoch: 834, batch: 1, loss: 3.3665590286254883, modularity_loss: -0.6086758971214294, purity_loss: 3.90051531791687, regularization_loss: 0.07471956312656403 # 11:37:46 --- INFO: epoch: 835, batch: 1, loss: 3.3664467334747314, modularity_loss: -0.6086772680282593, purity_loss: 3.900404691696167, regularization_loss: 0.07471925765275955 # 11:37:46 --- INFO: epoch: 836, batch: 1, loss: 3.366333484649658, modularity_loss: -0.6086784601211548, purity_loss: 3.9002928733825684, regularization_loss: 0.07471904903650284 # 11:37:46 --- INFO: epoch: 837, batch: 1, loss: 3.3662214279174805, modularity_loss: -0.6086785793304443, purity_loss: 3.9001808166503906, regularization_loss: 0.0747191533446312 # 11:37:47 --- INFO: epoch: 838, batch: 1, loss: 3.3661134243011475, modularity_loss: -0.6086772084236145, purity_loss: 3.900071144104004, regularization_loss: 0.07471956312656403 # 11:37:47 --- INFO: epoch: 839, batch: 1, loss: 3.3660027980804443, modularity_loss: -0.6086750030517578, purity_loss: 3.8999578952789307, regularization_loss: 0.0747198760509491 # 11:37:47 --- INFO: epoch: 840, batch: 1, loss: 3.365891456604004, modularity_loss: -0.6086729764938354, purity_loss: 3.8998444080352783, regularization_loss: 0.07471991330385208 # 11:37:48 --- INFO: epoch: 841, batch: 1, loss: 3.3657798767089844, modularity_loss: -0.6086714267730713, purity_loss: 3.899731397628784, regularization_loss: 0.07471980899572372 # 11:37:48 --- INFO: epoch: 842, batch: 1, loss: 3.3656721115112305, modularity_loss: -0.6086705923080444, purity_loss: 3.899623155593872, regularization_loss: 0.07471960783004761 # 11:37:48 --- INFO: epoch: 843, batch: 1, loss: 3.365560531616211, modularity_loss: -0.6086690425872803, purity_loss: 3.899510145187378, regularization_loss: 0.07471943646669388 # 11:37:49 --- INFO: epoch: 844, batch: 1, loss: 3.3654513359069824, modularity_loss: -0.6086664199829102, purity_loss: 3.8993983268737793, regularization_loss: 0.07471950352191925 # 11:37:49 --- INFO: epoch: 845, batch: 1, loss: 3.3653407096862793, modularity_loss: -0.6086632013320923, purity_loss: 3.8992841243743896, regularization_loss: 0.07471966743469238 # 11:37:49 --- INFO: epoch: 846, batch: 1, loss: 3.365234136581421, modularity_loss: -0.608659029006958, purity_loss: 3.8991734981536865, regularization_loss: 0.0747196301817894 # 11:37:50 --- INFO: epoch: 847, batch: 1, loss: 3.3651249408721924, modularity_loss: -0.6086568236351013, purity_loss: 3.899062156677246, regularization_loss: 0.0747196152806282 # 11:37:50 --- INFO: epoch: 848, batch: 1, loss: 3.365017890930176, modularity_loss: -0.6086575984954834, purity_loss: 3.898955821990967, regularization_loss: 0.07471958547830582 # 11:37:50 --- INFO: epoch: 849, batch: 1, loss: 3.3649096488952637, modularity_loss: -0.6086587905883789, purity_loss: 3.8988492488861084, regularization_loss: 0.07471923530101776 # 11:37:50 --- INFO: epoch: 850, batch: 1, loss: 3.364804744720459, modularity_loss: -0.6086594462394714, purity_loss: 3.89874529838562, regularization_loss: 0.07471877336502075 # 11:37:51 --- INFO: epoch: 851, batch: 1, loss: 3.3647005558013916, modularity_loss: -0.6086602210998535, purity_loss: 3.8986423015594482, regularization_loss: 0.07471855729818344 # 11:37:51 --- INFO: epoch: 852, batch: 1, loss: 3.3645946979522705, modularity_loss: -0.6086597442626953, purity_loss: 3.898535966873169, regularization_loss: 0.07471852749586105 # 11:37:51 --- INFO: epoch: 853, batch: 1, loss: 3.364490032196045, modularity_loss: -0.6086597442626953, purity_loss: 3.8984313011169434, regularization_loss: 0.07471845299005508 # 11:37:52 --- INFO: epoch: 854, batch: 1, loss: 3.3643879890441895, modularity_loss: -0.6086595058441162, purity_loss: 3.898329257965088, regularization_loss: 0.07471834868192673 # 11:37:52 --- INFO: epoch: 855, batch: 1, loss: 3.3642866611480713, modularity_loss: -0.6086603999137878, purity_loss: 3.898228645324707, regularization_loss: 0.07471837848424911 # 11:37:52 --- INFO: epoch: 856, batch: 1, loss: 3.3641834259033203, modularity_loss: -0.6086607575416565, purity_loss: 3.8981258869171143, regularization_loss: 0.0747184082865715 # 11:37:53 --- INFO: epoch: 857, batch: 1, loss: 3.3640835285186768, modularity_loss: -0.6086597442626953, purity_loss: 3.8980250358581543, regularization_loss: 0.07471821457147598 # 11:37:53 --- INFO: epoch: 858, batch: 1, loss: 3.3639824390411377, modularity_loss: -0.6086573600769043, purity_loss: 3.8979220390319824, regularization_loss: 0.07471783459186554 # 11:37:53 --- INFO: epoch: 859, batch: 1, loss: 3.363884449005127, modularity_loss: -0.6086558103561401, purity_loss: 3.897822618484497, regularization_loss: 0.07471772283315659 # 11:37:53 --- INFO: epoch: 860, batch: 1, loss: 3.363784074783325, modularity_loss: -0.6086548566818237, purity_loss: 3.897721290588379, regularization_loss: 0.07471761107444763 # 11:37:54 --- INFO: epoch: 861, batch: 1, loss: 3.3636832237243652, modularity_loss: -0.6086538434028625, purity_loss: 3.8976197242736816, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 862, batch: 1, loss: 3.363584518432617, modularity_loss: -0.6086530089378357, purity_loss: 3.897520065307617, regularization_loss: 0.07471734285354614 # 11:37:54 --- INFO: epoch: 863, batch: 1, loss: 3.363487720489502, modularity_loss: -0.6086521148681641, purity_loss: 3.8974220752716064, regularization_loss: 0.07471767067909241 # 11:37:55 --- INFO: epoch: 864, batch: 1, loss: 3.3633904457092285, modularity_loss: -0.6086492538452148, purity_loss: 3.897321939468384, regularization_loss: 0.07471783459186554 # 11:37:55 --- INFO: epoch: 865, batch: 1, loss: 3.363291025161743, modularity_loss: -0.6086451411247253, purity_loss: 3.8972184658050537, regularization_loss: 0.07471775263547897 # 11:37:55 --- INFO: epoch: 866, batch: 1, loss: 3.363197088241577, modularity_loss: -0.6086416244506836, purity_loss: 3.8971211910247803, regularization_loss: 0.07471756637096405 # 11:37:56 --- INFO: epoch: 867, batch: 1, loss: 3.3630990982055664, modularity_loss: -0.608640193939209, purity_loss: 3.897021770477295, regularization_loss: 0.07471758872270584 # 11:37:56 --- INFO: epoch: 868, batch: 1, loss: 3.363003969192505, modularity_loss: -0.608640193939209, purity_loss: 3.8969266414642334, regularization_loss: 0.07471749186515808 # 11:37:56 --- INFO: epoch: 869, batch: 1, loss: 3.3629074096679688, modularity_loss: -0.6086399555206299, purity_loss: 3.8968303203582764, regularization_loss: 0.07471711933612823 # 11:37:56 --- INFO: epoch: 870, batch: 1, loss: 3.362811803817749, modularity_loss: -0.6086399555206299, purity_loss: 3.8967349529266357, regularization_loss: 0.07471687346696854 # 11:37:57 --- INFO: epoch: 871, batch: 1, loss: 3.362717628479004, modularity_loss: -0.6086417436599731, purity_loss: 3.8966422080993652, regularization_loss: 0.07471714913845062 # 11:37:57 --- INFO: epoch: 872, batch: 1, loss: 3.362619638442993, modularity_loss: -0.6086426377296448, purity_loss: 3.896544933319092, regularization_loss: 0.07471726089715958 # 11:37:57 --- INFO: epoch: 873, batch: 1, loss: 3.362522602081299, modularity_loss: -0.6086419224739075, purity_loss: 3.8964476585388184, regularization_loss: 0.0747169554233551 # 11:37:58 --- INFO: epoch: 874, batch: 1, loss: 3.3624322414398193, modularity_loss: -0.6086419820785522, purity_loss: 3.896357536315918, regularization_loss: 0.07471676915884018 # 11:37:58 --- INFO: epoch: 875, batch: 1, loss: 3.362335205078125, modularity_loss: -0.608643651008606, purity_loss: 3.8962621688842773, regularization_loss: 0.07471683621406555 # 11:37:58 --- INFO: epoch: 876, batch: 1, loss: 3.3622403144836426, modularity_loss: -0.6086437702178955, purity_loss: 3.896167516708374, regularization_loss: 0.0747164860367775 # 11:37:58 --- INFO: epoch: 877, batch: 1, loss: 3.3621551990509033, modularity_loss: -0.6086429357528687, purity_loss: 3.8960821628570557, regularization_loss: 0.07471602410078049 # 11:37:59 --- INFO: epoch: 878, batch: 1, loss: 3.3620612621307373, modularity_loss: -0.6086430549621582, purity_loss: 3.8959882259368896, regularization_loss: 0.07471617311239243 # 11:37:59 --- INFO: epoch: 879, batch: 1, loss: 3.361969232559204, modularity_loss: -0.6086426973342896, purity_loss: 3.895895481109619, regularization_loss: 0.07471641898155212 # 11:37:59 --- INFO: epoch: 880, batch: 1, loss: 3.361877918243408, modularity_loss: -0.608640193939209, purity_loss: 3.8958020210266113, regularization_loss: 0.0747162401676178 # 11:38:00 --- INFO: epoch: 881, batch: 1, loss: 3.361787796020508, modularity_loss: -0.6086382865905762, purity_loss: 3.895709991455078, regularization_loss: 0.07471606880426407 # 11:38:00 --- INFO: epoch: 882, batch: 1, loss: 3.3616981506347656, modularity_loss: -0.6086363792419434, purity_loss: 3.895618438720703, regularization_loss: 0.07471617311239243 # 11:38:00 --- INFO: epoch: 883, batch: 1, loss: 3.3616058826446533, modularity_loss: -0.6086349487304688, purity_loss: 3.895524501800537, regularization_loss: 0.07471639662981033 # 11:38:01 --- INFO: epoch: 884, batch: 1, loss: 3.361515522003174, modularity_loss: -0.6086326837539673, purity_loss: 3.8954317569732666, regularization_loss: 0.07471631467342377 # 11:38:01 --- INFO: epoch: 885, batch: 1, loss: 3.3614261150360107, modularity_loss: -0.6086311340332031, purity_loss: 3.895340919494629, regularization_loss: 0.07471625506877899 # 11:38:01 --- INFO: epoch: 886, batch: 1, loss: 3.361335277557373, modularity_loss: -0.6086299419403076, purity_loss: 3.895249128341675, regularization_loss: 0.07471618801355362 # 11:38:01 --- INFO: epoch: 887, batch: 1, loss: 3.361248016357422, modularity_loss: -0.6086293458938599, purity_loss: 3.8951616287231445, regularization_loss: 0.07471571862697601 # 11:38:02 --- INFO: epoch: 888, batch: 1, loss: 3.36116099357605, modularity_loss: -0.6086305379867554, purity_loss: 3.895076274871826, regularization_loss: 0.07471520453691483 # 11:38:02 --- INFO: epoch: 889, batch: 1, loss: 3.361072540283203, modularity_loss: -0.6086328029632568, purity_loss: 3.8949902057647705, regularization_loss: 0.07471514493227005 # 11:38:02 --- INFO: epoch: 890, batch: 1, loss: 3.3609845638275146, modularity_loss: -0.6086337566375732, purity_loss: 3.8949031829833984, regularization_loss: 0.07471514493227005 # 11:38:03 --- INFO: epoch: 891, batch: 1, loss: 3.3608970642089844, modularity_loss: -0.6086312532424927, purity_loss: 3.894813299179077, regularization_loss: 0.0747150406241417 # 11:38:03 --- INFO: epoch: 892, batch: 1, loss: 3.3608102798461914, modularity_loss: -0.6086273193359375, purity_loss: 3.8947224617004395, regularization_loss: 0.07471522688865662 # 11:38:03 --- INFO: epoch: 893, batch: 1, loss: 3.3607234954833984, modularity_loss: -0.608623206615448, purity_loss: 3.8946311473846436, regularization_loss: 0.0747155025601387 # 11:38:04 --- INFO: epoch: 894, batch: 1, loss: 3.3606367111206055, modularity_loss: -0.6086186170578003, purity_loss: 3.8945398330688477, regularization_loss: 0.07471556216478348 # 11:38:04 --- INFO: epoch: 895, batch: 1, loss: 3.3605518341064453, modularity_loss: -0.6086139678955078, purity_loss: 3.8944501876831055, regularization_loss: 0.07471547275781631 # 11:38:04 --- INFO: epoch: 896, batch: 1, loss: 3.3604631423950195, modularity_loss: -0.6086108684539795, purity_loss: 3.8943583965301514, regularization_loss: 0.07471547275781631 # 11:38:05 --- INFO: epoch: 897, batch: 1, loss: 3.360379219055176, modularity_loss: -0.6086083054542542, purity_loss: 3.8942723274230957, regularization_loss: 0.07471532374620438 # 11:38:05 --- INFO: epoch: 898, batch: 1, loss: 3.360293388366699, modularity_loss: -0.608607292175293, purity_loss: 3.8941855430603027, regularization_loss: 0.0747152715921402 # 11:38:05 --- INFO: epoch: 899, batch: 1, loss: 3.3602099418640137, modularity_loss: -0.6086069345474243, purity_loss: 3.89410138130188, regularization_loss: 0.07471535354852676 # 11:38:06 --- INFO: epoch: 900, batch: 1, loss: 3.3601222038269043, modularity_loss: -0.6086060404777527, purity_loss: 3.8940131664276123, regularization_loss: 0.07471512258052826 # 11:38:06 --- INFO: epoch: 901, batch: 1, loss: 3.3600378036499023, modularity_loss: -0.6086047887802124, purity_loss: 3.893927812576294, regularization_loss: 0.07471483200788498 # 11:38:06 --- INFO: epoch: 902, batch: 1, loss: 3.359954595565796, modularity_loss: -0.608603835105896, purity_loss: 3.893843650817871, regularization_loss: 0.07471473515033722 # 11:38:07 --- INFO: epoch: 903, batch: 1, loss: 3.3598713874816895, modularity_loss: -0.6086021661758423, purity_loss: 3.893758773803711, regularization_loss: 0.07471466809511185 # 11:38:07 --- INFO: epoch: 904, batch: 1, loss: 3.3597888946533203, modularity_loss: -0.6085999011993408, purity_loss: 3.89367413520813, regularization_loss: 0.07471456378698349 # 11:38:07 --- INFO: epoch: 905, batch: 1, loss: 3.3597071170806885, modularity_loss: -0.6085981130599976, purity_loss: 3.8935906887054443, regularization_loss: 0.07471449673175812 # 11:38:07 --- INFO: epoch: 906, batch: 1, loss: 3.3596251010894775, modularity_loss: -0.6085958480834961, purity_loss: 3.8935065269470215, regularization_loss: 0.07471442967653275 # 11:38:08 --- INFO: epoch: 907, batch: 1, loss: 3.3595428466796875, modularity_loss: -0.6085939407348633, purity_loss: 3.8934223651885986, regularization_loss: 0.07471437752246857 # 11:38:08 --- INFO: epoch: 908, batch: 1, loss: 3.3594632148742676, modularity_loss: -0.6085922122001648, purity_loss: 3.893341064453125, regularization_loss: 0.07471431791782379 # 11:38:08 --- INFO: epoch: 909, batch: 1, loss: 3.359382390975952, modularity_loss: -0.6085900068283081, purity_loss: 3.8932583332061768, regularization_loss: 0.07471413165330887 # 11:38:09 --- INFO: epoch: 910, batch: 1, loss: 3.3593056201934814, modularity_loss: -0.6085877418518066, purity_loss: 3.893179416656494, regularization_loss: 0.07471396774053574 # 11:38:09 --- INFO: epoch: 911, batch: 1, loss: 3.3592257499694824, modularity_loss: -0.6085848808288574, purity_loss: 3.893096685409546, regularization_loss: 0.07471387088298798 # 11:38:09 --- INFO: epoch: 912, batch: 1, loss: 3.359145402908325, modularity_loss: -0.6085816621780396, purity_loss: 3.8930132389068604, regularization_loss: 0.0747138261795044 # 11:38:10 --- INFO: epoch: 913, batch: 1, loss: 3.359071731567383, modularity_loss: -0.6085776090621948, purity_loss: 3.8929355144500732, regularization_loss: 0.0747138038277626 # 11:38:10 --- INFO: epoch: 914, batch: 1, loss: 3.3589890003204346, modularity_loss: -0.6085737943649292, purity_loss: 3.8928489685058594, regularization_loss: 0.07471388578414917 # 11:38:10 --- INFO: epoch: 915, batch: 1, loss: 3.3589136600494385, modularity_loss: -0.6085696220397949, purity_loss: 3.8927693367004395, regularization_loss: 0.07471396774053574 # 11:38:10 --- INFO: epoch: 916, batch: 1, loss: 3.358834743499756, modularity_loss: -0.6085658073425293, purity_loss: 3.892686605453491, regularization_loss: 0.07471392303705215 # 11:38:11 --- INFO: epoch: 917, batch: 1, loss: 3.3587582111358643, modularity_loss: -0.608563244342804, purity_loss: 3.8926076889038086, regularization_loss: 0.07471374422311783 # 11:38:11 --- INFO: epoch: 918, batch: 1, loss: 3.358682632446289, modularity_loss: -0.6085621118545532, purity_loss: 3.892531156539917, regularization_loss: 0.07471358776092529 # 11:38:11 --- INFO: epoch: 919, batch: 1, loss: 3.3586058616638184, modularity_loss: -0.6085608005523682, purity_loss: 3.89245343208313, regularization_loss: 0.07471328228712082 # 11:38:12 --- INFO: epoch: 920, batch: 1, loss: 3.358531951904297, modularity_loss: -0.6085584163665771, purity_loss: 3.8923773765563965, regularization_loss: 0.07471304386854172 # 11:38:12 --- INFO: epoch: 921, batch: 1, loss: 3.358454704284668, modularity_loss: -0.6085555553436279, purity_loss: 3.8922972679138184, regularization_loss: 0.07471306622028351 # 11:38:12 --- INFO: epoch: 922, batch: 1, loss: 3.358382225036621, modularity_loss: -0.608552098274231, purity_loss: 3.892221212387085, regularization_loss: 0.07471314072608948 # 11:38:13 --- INFO: epoch: 923, batch: 1, loss: 3.358306884765625, modularity_loss: -0.6085484027862549, purity_loss: 3.8921422958374023, regularization_loss: 0.07471313327550888 # 11:38:13 --- INFO: epoch: 924, batch: 1, loss: 3.3582303524017334, modularity_loss: -0.60854572057724, purity_loss: 3.8920629024505615, regularization_loss: 0.07471310347318649 # 11:38:13 --- INFO: epoch: 925, batch: 1, loss: 3.358159303665161, modularity_loss: -0.6085429191589355, purity_loss: 3.891989231109619, regularization_loss: 0.07471305876970291 # 11:38:13 --- INFO: epoch: 926, batch: 1, loss: 3.3580844402313232, modularity_loss: -0.6085397005081177, purity_loss: 3.891911268234253, regularization_loss: 0.07471288740634918 # 11:38:14 --- INFO: epoch: 927, batch: 1, loss: 3.358010768890381, modularity_loss: -0.6085366010665894, purity_loss: 3.8918347358703613, regularization_loss: 0.07471274584531784 # 11:38:14 --- INFO: epoch: 928, batch: 1, loss: 3.3579370975494385, modularity_loss: -0.6085345149040222, purity_loss: 3.891758918762207, regularization_loss: 0.07471276074647903 # 11:38:14 --- INFO: epoch: 929, batch: 1, loss: 3.3578662872314453, modularity_loss: -0.6085318922996521, purity_loss: 3.8916854858398438, regularization_loss: 0.07471269369125366 # 11:38:15 --- INFO: epoch: 930, batch: 1, loss: 3.35779070854187, modularity_loss: -0.6085291504859924, purity_loss: 3.8916072845458984, regularization_loss: 0.0747125968337059 # 11:38:15 --- INFO: epoch: 931, batch: 1, loss: 3.3577191829681396, modularity_loss: -0.6085257530212402, purity_loss: 3.8915321826934814, regularization_loss: 0.07471267879009247 # 11:38:15 --- INFO: epoch: 932, batch: 1, loss: 3.35764741897583, modularity_loss: -0.6085220575332642, purity_loss: 3.8914568424224854, regularization_loss: 0.07471269369125366 # 11:38:16 --- INFO: epoch: 933, batch: 1, loss: 3.357576847076416, modularity_loss: -0.6085176467895508, purity_loss: 3.8913819789886475, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 934, batch: 1, loss: 3.357503890991211, modularity_loss: -0.6085143089294434, purity_loss: 3.891305685043335, regularization_loss: 0.07471262663602829 # 11:38:16 --- INFO: epoch: 935, batch: 1, loss: 3.3574321269989014, modularity_loss: -0.6085120439529419, purity_loss: 3.8912315368652344, regularization_loss: 0.07471258193254471 # 11:38:17 --- INFO: epoch: 936, batch: 1, loss: 3.35736083984375, modularity_loss: -0.6085104942321777, purity_loss: 3.8911592960357666, regularization_loss: 0.07471216470003128 # 11:38:17 --- INFO: epoch: 937, batch: 1, loss: 3.3572921752929688, modularity_loss: -0.6085103750228882, purity_loss: 3.8910906314849854, regularization_loss: 0.07471182942390442 # 11:38:17 --- INFO: epoch: 938, batch: 1, loss: 3.3572206497192383, modularity_loss: -0.6085109114646912, purity_loss: 3.891019821166992, regularization_loss: 0.0747116357088089 # 11:38:17 --- INFO: epoch: 939, batch: 1, loss: 3.3571510314941406, modularity_loss: -0.6085106730461121, purity_loss: 3.8909502029418945, regularization_loss: 0.0747113898396492 # 11:38:18 --- INFO: epoch: 940, batch: 1, loss: 3.3570804595947266, modularity_loss: -0.6085097193717957, purity_loss: 3.890878677368164, regularization_loss: 0.07471135258674622 # 11:38:18 --- INFO: epoch: 941, batch: 1, loss: 3.3570122718811035, modularity_loss: -0.6085082292556763, purity_loss: 3.8908090591430664, regularization_loss: 0.07471150159835815 # 11:38:18 --- INFO: epoch: 942, batch: 1, loss: 3.356943130493164, modularity_loss: -0.608505129814148, purity_loss: 3.8907368183135986, regularization_loss: 0.07471148669719696 # 11:38:19 --- INFO: epoch: 943, batch: 1, loss: 3.3568732738494873, modularity_loss: -0.6085014939308167, purity_loss: 3.8906633853912354, regularization_loss: 0.0747114047408104 # 11:38:19 --- INFO: epoch: 944, batch: 1, loss: 3.3568053245544434, modularity_loss: -0.6084985733032227, purity_loss: 3.890592336654663, regularization_loss: 0.07471145689487457 # 11:38:19 --- INFO: epoch: 945, batch: 1, loss: 3.3567354679107666, modularity_loss: -0.6084975600242615, purity_loss: 3.89052152633667, regularization_loss: 0.07471141964197159 # 11:38:20 --- INFO: epoch: 946, batch: 1, loss: 3.3566694259643555, modularity_loss: -0.6084966659545898, purity_loss: 3.8904547691345215, regularization_loss: 0.07471121847629547 # 11:38:20 --- INFO: epoch: 947, batch: 1, loss: 3.3566014766693115, modularity_loss: -0.6084957122802734, purity_loss: 3.8903861045837402, regularization_loss: 0.0747111588716507 # 11:38:20 --- INFO: epoch: 948, batch: 1, loss: 3.3565309047698975, modularity_loss: -0.6084946393966675, purity_loss: 3.8903143405914307, regularization_loss: 0.07471119612455368 # 11:38:20 --- INFO: epoch: 949, batch: 1, loss: 3.3564648628234863, modularity_loss: -0.6084920167922974, purity_loss: 3.8902456760406494, regularization_loss: 0.07471106946468353 # 11:38:21 --- INFO: epoch: 950, batch: 1, loss: 3.3563952445983887, modularity_loss: -0.6084898114204407, purity_loss: 3.89017391204834, regularization_loss: 0.07471103221178055 # 11:38:21 --- INFO: epoch: 951, batch: 1, loss: 3.3563313484191895, modularity_loss: -0.6084879636764526, purity_loss: 3.890108346939087, regularization_loss: 0.07471106201410294 # 11:38:21 --- INFO: epoch: 952, batch: 1, loss: 3.356266975402832, modularity_loss: -0.6084863543510437, purity_loss: 3.890042304992676, regularization_loss: 0.074710913002491 # 11:38:22 --- INFO: epoch: 953, batch: 1, loss: 3.356199264526367, modularity_loss: -0.6084855794906616, purity_loss: 3.8899741172790527, regularization_loss: 0.07471081614494324 # 11:38:22 --- INFO: epoch: 954, batch: 1, loss: 3.356135845184326, modularity_loss: -0.6084851026535034, purity_loss: 3.8899102210998535, regularization_loss: 0.07471080124378204 # 11:38:22 --- INFO: epoch: 955, batch: 1, loss: 3.3560678958892822, modularity_loss: -0.6084853410720825, purity_loss: 3.8898427486419678, regularization_loss: 0.07471051812171936 # 11:38:23 --- INFO: epoch: 956, batch: 1, loss: 3.356001377105713, modularity_loss: -0.6084868907928467, purity_loss: 3.8897781372070312, regularization_loss: 0.0747101902961731 # 11:38:23 --- INFO: epoch: 957, batch: 1, loss: 3.3559372425079346, modularity_loss: -0.6084891557693481, purity_loss: 3.889716386795044, regularization_loss: 0.074709951877594 # 11:38:23 --- INFO: epoch: 958, batch: 1, loss: 3.3558707237243652, modularity_loss: -0.6084906458854675, purity_loss: 3.8896515369415283, regularization_loss: 0.07470972836017609 # 11:38:24 --- INFO: epoch: 959, batch: 1, loss: 3.355807304382324, modularity_loss: -0.6084908246994019, purity_loss: 3.8895885944366455, regularization_loss: 0.07470959424972534 # 11:38:24 --- INFO: epoch: 960, batch: 1, loss: 3.355739116668701, modularity_loss: -0.6084907054901123, purity_loss: 3.8895201683044434, regularization_loss: 0.07470975816249847 # 11:38:24 --- INFO: epoch: 961, batch: 1, loss: 3.3556771278381348, modularity_loss: -0.6084891557693481, purity_loss: 3.889456272125244, regularization_loss: 0.07470987737178802 # 11:38:25 --- INFO: epoch: 962, batch: 1, loss: 3.3556151390075684, modularity_loss: -0.6084870100021362, purity_loss: 3.889392375946045, regularization_loss: 0.07470982521772385 # 11:38:25 --- INFO: epoch: 963, batch: 1, loss: 3.35555362701416, modularity_loss: -0.608485221862793, purity_loss: 3.889328956604004, regularization_loss: 0.07470980286598206 # 11:38:25 --- INFO: epoch: 964, batch: 1, loss: 3.3554887771606445, modularity_loss: -0.6084840893745422, purity_loss: 3.889263153076172, regularization_loss: 0.07470960915088654 # 11:38:26 --- INFO: epoch: 965, batch: 1, loss: 3.355426549911499, modularity_loss: -0.6084831953048706, purity_loss: 3.889200448989868, regularization_loss: 0.07470928132534027 # 11:38:26 --- INFO: epoch: 966, batch: 1, loss: 3.355362892150879, modularity_loss: -0.6084827184677124, purity_loss: 3.88913631439209, regularization_loss: 0.07470914721488953 # 11:38:26 --- INFO: epoch: 967, batch: 1, loss: 3.3552968502044678, modularity_loss: -0.6084824204444885, purity_loss: 3.8890700340270996, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 968, batch: 1, loss: 3.355233907699585, modularity_loss: -0.6084803938865662, purity_loss: 3.889005184173584, regularization_loss: 0.07470910996198654 # 11:38:27 --- INFO: epoch: 969, batch: 1, loss: 3.355172634124756, modularity_loss: -0.6084768772125244, purity_loss: 3.8889403343200684, regularization_loss: 0.07470916956663132 # 11:38:27 --- INFO: epoch: 970, batch: 1, loss: 3.355111837387085, modularity_loss: -0.6084741353988647, purity_loss: 3.8888766765594482, regularization_loss: 0.07470931112766266 # 11:38:28 --- INFO: epoch: 971, batch: 1, loss: 3.3550498485565186, modularity_loss: -0.6084709167480469, purity_loss: 3.8888115882873535, regularization_loss: 0.07470913976430893 # 11:38:28 --- INFO: epoch: 972, batch: 1, loss: 3.3549880981445312, modularity_loss: -0.6084688901901245, purity_loss: 3.8887481689453125, regularization_loss: 0.07470890879631042 # 11:38:28 --- INFO: epoch: 973, batch: 1, loss: 3.3549273014068604, modularity_loss: -0.6084681153297424, purity_loss: 3.8886866569519043, regularization_loss: 0.07470883429050446 # 11:38:29 --- INFO: epoch: 974, batch: 1, loss: 3.354865550994873, modularity_loss: -0.6084681749343872, purity_loss: 3.888624906539917, regularization_loss: 0.07470870018005371 # 11:38:29 --- INFO: epoch: 975, batch: 1, loss: 3.3548054695129395, modularity_loss: -0.608467698097229, purity_loss: 3.8885645866394043, regularization_loss: 0.07470859587192535 # 11:38:29 --- INFO: epoch: 976, batch: 1, loss: 3.354745626449585, modularity_loss: -0.6084665060043335, purity_loss: 3.8885035514831543, regularization_loss: 0.07470859587192535 # 11:38:30 --- INFO: epoch: 977, batch: 1, loss: 3.3546838760375977, modularity_loss: -0.6084654331207275, purity_loss: 3.8884408473968506, regularization_loss: 0.07470858097076416 # 11:38:30 --- INFO: epoch: 978, batch: 1, loss: 3.3546218872070312, modularity_loss: -0.6084632873535156, purity_loss: 3.8883767127990723, regularization_loss: 0.07470840215682983 # 11:38:30 --- INFO: epoch: 979, batch: 1, loss: 3.3545637130737305, modularity_loss: -0.6084608435630798, purity_loss: 3.8883161544799805, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 980, batch: 1, loss: 3.3545026779174805, modularity_loss: -0.6084582805633545, purity_loss: 3.8882527351379395, regularization_loss: 0.07470832020044327 # 11:38:31 --- INFO: epoch: 981, batch: 1, loss: 3.3544421195983887, modularity_loss: -0.6084554195404053, purity_loss: 3.8881893157958984, regularization_loss: 0.07470821589231491 # 11:38:31 --- INFO: epoch: 982, batch: 1, loss: 3.354381799697876, modularity_loss: -0.6084536910057068, purity_loss: 3.888127565383911, regularization_loss: 0.07470792531967163 # 11:38:32 --- INFO: epoch: 983, batch: 1, loss: 3.3543262481689453, modularity_loss: -0.6084543466567993, purity_loss: 3.8880727291107178, regularization_loss: 0.0747077539563179 # 11:38:32 --- INFO: epoch: 984, batch: 1, loss: 3.3542652130126953, modularity_loss: -0.6084551811218262, purity_loss: 3.8880128860473633, regularization_loss: 0.07470749318599701 # 11:38:32 --- INFO: epoch: 985, batch: 1, loss: 3.3542048931121826, modularity_loss: -0.6084557771682739, purity_loss: 3.887953519821167, regularization_loss: 0.07470718771219254 # 11:38:32 --- INFO: epoch: 986, batch: 1, loss: 3.354147434234619, modularity_loss: -0.6084555387496948, purity_loss: 3.8878958225250244, regularization_loss: 0.07470723986625671 # 11:38:33 --- INFO: epoch: 987, batch: 1, loss: 3.3540899753570557, modularity_loss: -0.6084539890289307, purity_loss: 3.8878366947174072, regularization_loss: 0.07470730692148209 # 11:38:33 --- INFO: epoch: 988, batch: 1, loss: 3.3540306091308594, modularity_loss: -0.6084518432617188, purity_loss: 3.887775182723999, regularization_loss: 0.07470717281103134 # 11:38:33 --- INFO: epoch: 989, batch: 1, loss: 3.3539719581604004, modularity_loss: -0.6084514856338501, purity_loss: 3.887716293334961, regularization_loss: 0.07470714300870895 # 11:38:34 --- INFO: epoch: 990, batch: 1, loss: 3.353915214538574, modularity_loss: -0.608452558517456, purity_loss: 3.887660503387451, regularization_loss: 0.07470720261335373 # 11:38:34 --- INFO: epoch: 991, batch: 1, loss: 3.3538575172424316, modularity_loss: -0.6084526777267456, purity_loss: 3.8876030445098877, regularization_loss: 0.07470700889825821 # 11:38:34 --- INFO: epoch: 992, batch: 1, loss: 3.3538012504577637, modularity_loss: -0.6084530353546143, purity_loss: 3.887547492980957, regularization_loss: 0.0747067853808403 # 11:38:35 --- INFO: epoch: 993, batch: 1, loss: 3.3537447452545166, modularity_loss: -0.6084531545639038, purity_loss: 3.887491226196289, regularization_loss: 0.07470668852329254 # 11:38:35 --- INFO: epoch: 994, batch: 1, loss: 3.353687286376953, modularity_loss: -0.6084524393081665, purity_loss: 3.8874335289001465, regularization_loss: 0.0747063159942627 # 11:38:35 --- INFO: epoch: 995, batch: 1, loss: 3.3536300659179688, modularity_loss: -0.6084518432617188, purity_loss: 3.887375831604004, regularization_loss: 0.07470611482858658 # 11:38:36 --- INFO: epoch: 996, batch: 1, loss: 3.353571891784668, modularity_loss: -0.6084519624710083, purity_loss: 3.887317657470703, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 997, batch: 1, loss: 3.353517770767212, modularity_loss: -0.608451247215271, purity_loss: 3.8872628211975098, regularization_loss: 0.07470627874135971 # 11:38:36 --- INFO: epoch: 998, batch: 1, loss: 3.353461265563965, modularity_loss: -0.6084499955177307, purity_loss: 3.887204885482788, regularization_loss: 0.07470636069774628 # 11:38:37 --- INFO: epoch: 999, batch: 1, loss: 3.3534069061279297, modularity_loss: -0.6084499359130859, purity_loss: 3.887150526046753, regularization_loss: 0.07470645755529404 # 11:38:37 --- INFO: epoch: 1000, batch: 1, loss: 3.3533525466918945, modularity_loss: -0.6084506511688232, purity_loss: 3.887096881866455, regularization_loss: 0.07470621913671494 # 11:38:37 --- INFO: Training process end. # 11:38:37 --- INFO: Evaluating process start. # 11:38:37 --- INFO: Evaluate loss, {'modularity_loss': -0.6084526777267456, 'purity_loss': 3.8870439529418945, 'regularization_loss': 0.07470588386058807, 'total_loss': 3.353297233581543} # 11:38:37 --- INFO: Evaluating process end. # 11:38:37 --- INFO: Predicting process start. # 11:38:37 --- INFO: Generating prediction results for mouse2_slice229. # 11:38:38 --- INFO: Generating prediction results for mouse2_slice300. # 11:38:38 --- INFO: Predicting process end. # 11:38:38 --- INFO: --------------------- GNN end ---------------------- # 11:38:38 --- INFO: ----------------- Niche trajectory ------------------ # 11:38:38 --- INFO: Loading consolidate s_array and out_adj_array... # 11:38:38 --- INFO: Loading dataset. # 11:38:38 --- INFO: Maximum number of cell in one sample is: 5100. # Processing... # 11:38:38 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # Done! # 11:38:39 --- INFO: Processing sample 1 of 2: mouse2_slice229 # 11:38:39 --- INFO: Processing sample 2 of 2: mouse2_slice300 # 11:38:39 --- INFO: Calculating NTScore for each niche. # 11:38:39 --- INFO: Finding niche trajectory with maximum connectivity using Brute Force. # 11:38:39 --- INFO: Calculating NTScore for each niche cluster based on the trajectory path. # 11:38:39 --- INFO: Projecting NTScore from niche-level to cell-level. # 11:38:39 --- INFO: Output NTScore tables. # 11:38:39 --- INFO: --------------- Niche trajectory end ---------------- 16.1.3 visualization 16.1.3.1 load ONTraC results giotto_obj <- loadOntraCResults(gobject = giotto_obj, ontrac_results_dir = data_path) The NTScore and binarized niche cluster info were stored in cell metadata head(pDataDT(giotto_obj, spat_unit = "cell", feat_type = "niche cluster")) # cell_ID sample_id slice_id class_label subclass label list_ID NicheCluster NTScore # <char> <char> <char> <char> <char> <char> <char> <int> <num> # 1: mouse2_slice229-100101435705986292663283283043431511315 mouse2_sample6 mouse2_slice229 Glutamatergic L6 CT L6_CT_5 mouse2_slice229 2 0.7998957 # 2: mouse2_slice229-100104370212612969023746137269354247741 mouse2_sample6 mouse2_slice229 Other OPC OPC mouse2_slice229 0 0.2003027 # 3: mouse2_slice229-100128078183217482733448056590230529739 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2350597 # 4: mouse2_slice229-100209662400867003194056898065587980841 mouse2_sample6 mouse2_slice229 Other Oligo Oligo_1 mouse2_slice229 1 0.3990417 # 5: mouse2_slice229-100218038012295593766653119076639444055 mouse2_sample6 mouse2_slice229 Glutamatergic L2/3 IT L23_IT_4 mouse2_slice229 0 0.2910255 # 6: mouse2_slice229-100252992997994275968450436343196667192 mouse2_sample6 mouse2_slice229 Other Astro Astro_2 mouse2_slice229 2 0.8007477 The probability matrix of each cell assigned to each niche cluster and connectivity between niche cluster were stored here. GiottoClass::list_expression(giotto_obj) # spat_unit feat_type name # <char> <char> <char> # 1: cell rna raw # 2: cell niche cluster prob # 3: niche cluster connectivity normalized 16.1.3.2 niche cluster prob spatFeatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'niche cluster', expression_values = "prob", group_by = 'list_ID', feats = rownames(giotto_obj@expression$cell$`niche cluster`$prob), point_border_col = 'gray' ) 16.1.3.3 binarized niche cluster spatPlot2D(giotto_obj, spat_unit = "cell", group_by = 'list_ID', cell_color = "NicheCluster", color_as_factor = T, point_size = 1, point_border_stroke = NA, title="Niche cluster") 16.1.3.4 niche cluster spatial connectivity set.seed(100) # fix the node positions plotNicheClusterConnectivity(gobject = giotto_obj) 16.1.3.5 NT score spatPlot2D(gobject = giotto_obj, spat_unit = 'cell', feat_type = 'rna', group_by = 'list_ID', cell_color = "NTScore", color_as_factor = FALSE, cell_color_gradient = "turbo", point_size = 1, point_border_stroke = NA ) plotCellTypeNTScore(gobject = giotto_obj, cell_type = "subclass") 16.1.3.6 Cell type composition within niche cluster plotCTCompositionInNicheCluster(gobject = giotto_obj, cell_type = "subclass") "],["interactivity-with-the-rspatial-ecosystem.html", "17 Interactivity with the R/Spatial ecosystem 17.1 Kriging 17.2 Downloading the dataset 17.3 Importing visium data 17.4 Adding cell polygons to Giotto object 17.5 Performing kriging 17.6 Reading in larger dataset 17.7 Analyzing interpolated features", " 17 Interactivity with the R/Spatial ecosystem Jeff Sheridan August 7th 2024 17.1 Kriging Low resolution spatial data typically covers multiple cells making it difficult to delineate the cell contribution to gene expression. Using a process called kriging we can interpolate gene expression at the single cell levels from low resolution datasets. Kriging is a spatial interpolation technique that estimates unknown values at specific locations by weighing nearby known values based on distance and spatial trends. It uses a model to account for both the distance between points and the overall pattern in the data to make accurate predictions. By taking discrete measurement spots, such as those used for visium, we can interpolate gene expression to a finer scale using kriging. 17.1.1 Visium technology Figure 17.1: Overview of Visium. Source: 10X Genomics. Visium by 10x Genomics is a spatial gene expression platform that allows for the mapping of gene expression to high-resolution histology through RNA sequencing The process involves placing a tissue section on a specially prepared slide with an array of barcoded spots, which are 55 µm in diameter with a spot to spot distance of 100 µm. Each spot contains unique barcodes that capture the mRNA from the tissue section, preserving the spatial information. After the tissue is imaged and RNA is captured, the mRNA is sequenced, and the data is mapped back to the tissue’s spatial coordinates. This technology is particularly useful in understanding complex tissue environments, such as tumors, by providing insights into how gene expression varies across different regions. 17.1.2 Dataset For this tutorial we’ll be using the mouse brain dataset described in section 6. Visium datasets require a high resolution H&E or IF image to align spots to. Using these images we can identify individual nuclei and cells to be used for kriging. Identifying nuclei is outside the scope of the current tutorial but is required to perform kriging. 17.1.3 Generating a geojson file of nuclei location For the following sections we will need to create a geojson that contains polygon information for the nuclei in the sample. We will be providing this in the following link, however when using for your own datasets this will need to be done outside of Giotto. A tutorial for this using qupath can be found here. 17.2 Downloading the dataset We first need to import a dataset that we want to perform kriging on. data_directory <- "data" download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz") download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz", destfile = "data/V1_Adult_Mouse_Brain_spatial.tar.gz") After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/”. 17.3 Importing visium data We’re going to begin by creating a Giotto object for the visium mouse brain dataset. This tutorial won’t go into detail about each of these steps as these have been covered for this dataset in section 6. To get the best results when performing gene expression interpolation we need to identify spatially distinct genes. Therefore, we need to perform nearest neighbour to create a spatial network. library(Giotto) save_directory <- 'results' visium_save_directory <- file.path(save_directory, 'visium_mouse_brain') subcell_save_directory <- file.path(save_directory, 'pseudo_subcellular/') instrs <- createGiottoInstructions(show_plot = TRUE, save_plot = TRUE, save_dir = visium_save_directory) v_brain <- createGiottoVisiumObject(data_directory, gene_column_index = 2, instructions = instrs) # Subset to in tissue only cm = pDataDT(v_brain) in_tissue_barcodes = cm[in_tissue == 1]$cell_ID v_brain = subsetGiotto(v_brain, cell_ids = in_tissue_barcodes) # Filter v_brain = filterGiotto(gobject = v_brain, expression_threshold = 1, feat_det_in_min_cells = 50, min_det_feats_per_cell = 1000, expression_values = c('raw')) # Normalize v_brain = normalizeGiotto(gobject = v_brain, scalefactor = 6000, verbose = TRUE) # Add stats v_brain = addStatistics(gobject = v_brain) # ID HVF v_brain = calculateHVF(gobject = v_brain, method = "cov_loess") fm = fDataDT(v_brain) hv_feats = fm[hvf == 'yes' & perc_cells > 3 & mean_expr_det > 0.4]$feat_ID length(hv_feats) # Dimension Reductions v_brain = runPCA(gobject = v_brain, feats_to_use = hv_feats) v_brain = runUMAP(v_brain, dimensions_to_use = 1:10, n_neighbors = 15, set_seed = TRUE) # NN Network v_brain = createNearestNetwork(gobject = v_brain, dimensions_to_use = 1:10, k = 15) # Leiden Cluster v_brain = doLeidenCluster(gobject = v_brain, resolution = 0.4, n_iterations = 1000, set_seed = TRUE) # Spatial Network (kNN) v_brain <- createSpatialNetwork(gobject = v_brain, method = 'kNN', k = 5, maximum_distance_knn = 400, name = 'spatial_network') spatPlot2D(gobject = v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 1.5, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = T, save_param = list(save_name = '03_ses6_1_vis_spat')) Here we can see the clustering of the regular visium spots is able to identify distinct regions of the mouse brain. Figure 17.2: Mouse brain spatial plot showing leiden clustering 17.3.1 Identifying spatially organized features We need to identify genes to be used for interpolation. This works best with genes that are spatially distinct. To identify these genes we’ll use binSpect(). For this tutorial we’ll only use the top 15 spatially distinct genes. The more genes used for interpolation the longer the analysis will take. When running this for your own datasets you should use more genes. We are only using 15 here to minimize analysis time. # Spatially Variable Features ranktest = binSpect(v_brain, bin_method = 'rank', calc_hub = T, hub_min_int = 5, spatial_network_name = 'spatial_network', do_parallel = T, cores = 8) #not able to provide a seed number, so do not set one # Getting the top 15 spatially organized genes ext_spatial_features = ranktest[1:15,]$feats 17.4 Adding cell polygons to Giotto object 17.4.1 Read in the poly information First we need to read in the geojson file that contains the cell polygons that we’ll interpolate gene expression onto. These will then be added to the Giotto object as a new polygon object. This won’t affect the visium polygons. Both polygons will be stored within the same Giotto object. # Read in the data stardist_cell_poly_path <- file.path(data_directory, "segmentations/stardist_only_cell_bounds.geojson") stardist_cell_gpoly <- createGiottoPolygonsFromGeoJSON(GeoJSON = stardist_cell_poly_path, name = "stardist_cell", calc_centroids = TRUE) stardist_cell_gpoly <- flip(stardist_cell_gpoly) 17.4.2 Vizualizing polygons Below we can see a vizualization of the polygons for the visium and the nuclei we identified from the H&E image. The visium dataset has 2698 spots compared to teh 36694 nuclei we identified. Just using the visium spots we’re therefore losing a lot of the spatial data for individual cells. With the increased number of spots and them directly correlating with the tissue, through the spots alone we are able to better see the actual sturcture of the mouse brain. plot(getPolygonInfo(v_brain)) plot(stardist_cell_gpoly, max_poly = 1e6) Figure 17.3: Mouse brain cell polygons from the visium dataset Figure 17.4: Mouse brain cell polygons with artifacts removed and flipped 17.4.3 Showing Giotto object prior to polygon addition Before we add the polygons we’re can see the gobject contains ‘cell’ as a spatial unit and a polygon. print(v_brain) Figure 17.5: Giotto object before adding subcellular polygons. 17.4.4 Adding polygons to giotto object After we add the nuclei polygons we can see that a new polygon name, ‘stardist_cell’ has been added to the gobject. v_brain <- addGiottoPolygons(v_brain, gpolygons = list('stardist_cell' = stardist_cell_gpoly)) print(v_brain) Figure 17.6: Giotto object after to adding subcellular polygons. 17.4.5 Check polygon information We can now see the addition of the new polygons under the name ‘stardist_cell’. Each of the new polyons is given a unique poly_ID as shown below. Each polygon is also added into same space as the original visium spots, therefore line up with the same image as the visium spots. poly_info <- getPolygonInfo(v_brain,polygon_name = 'stardist_cell') print(poly_info) Figure 17.7: Polygon information for stardist_cell. 17.5 Performing kriging 17.5.1 Interpolating features Now we can perform the first step in interpolating expression data onto cell polygons. This involves creating a raster image for the gene expression of each of the selected genes. The steps from here can be time consuming and require large amounts of memory. We will only be analyzing 15 genes to show the process of expression interpolation. For clustering and other analyses more genes are required. future::plan(future::multisession()) # comment out for single threading subcell_v_brain <- interpolateFeature(v_brain, spat_unit = 'cell', feat_type = 'rna', ext = ext(v_brain), feats = ext_spatial_features, overwrite = T) print(subcell_v_brain) Figure 17.8: Giotto object after to interpolating features. Addition of images for each interoplated feature (left) and an example of rasterized gene expression image (right). For each gene that we interpolate a raster image is exported based on the gene expression. Shown below is an example of an output for the gene Pantr1. Figure 17.9: Raster of gene expression interpolation for Pantr1 17.5.2 Expression overlap The raster we created above gives the gene expression in a graphical form. We next need to determine how that relates to the nuclei location. To determine that we will calculate the overlap of the rasterized gene expression image to the polygons supplied earlier. This step also takes more time the more genes that are provided. For large datasets please allow up to multiple hours for these steps to run. subcell_v_brain <- calculateOverlapPolygonImages(gobject = subcell_v_brain, name_overlap = "rna", spatial_info = "stardist_cell", image_names = ext_spatial_features) subcell_v_brain <- Giotto::overlapToMatrix(x = subcell_v_brain, poly_info = "stardist_cell", feat_info = "rna", aggr_function = "sum", type='intensity') After performing the overlap we now have expression data for each gene provided. This can be seen below where we see the interpolated gene expression for genes in each of the nuclei we identified. Figure 17.10: Gene expression for cells based on interpolation. 17.6 Reading in larger dataset For better results more genes are required. The above data used only 15 genes. We will now read in a dataset that has 1500 interpolated genes an use this for the remained of the tutorial. If you haven’t downloaded this dataset please download it here. subcell_v_brain <- loadGiotto(file.path(data_directory, 'subcellular_gobject')) 17.7 Analyzing interpolated features 17.7.1 Filter and normalization Now that we have a valid spat unit and gene expression data for each of the provided genes we can now perform the same analyses we used for the regular visium data. Please note that due to the differences in cell number that the valeus used for the current analysis aren’t identical to the visium analysis. subcell_v_brain <- filterGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "raw", expression_threshold = 1, feat_det_in_min_cells = 0, min_det_feats_per_cell = 1) subcell_v_brain <- normalizeGiotto(gobject = subcell_v_brain, spat_unit = "stardist_cell", scalefactor = 6000, verbose = T) 17.7.2 Visualizing gene expression from interpolated expression Since we have the gene expression information for both the visium and the interpolated gene expression we can vizualize gene expression for both from the same Giotto object. We will look at the expression for two genes ‘Sparc’ and ‘Pantr1’ for both the visium and interpolated data. spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Sparc', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_sparc')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'cell', gradient_style = 'sequential', feats = 'Pantr1', cell_color_gradient = 'Geyser', point_size = 2, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1_vis')) spatFeatPlot2D(subcell_v_brain, spat_unit = 'stardist_cell', gradient_style = 'sequential', cell_color_gradient = 'Geyser', feats = 'Pantr1', point_size = 0.6, save_plot = TRUE, show_image = TRUE, save_param = list(save_name = '03_ses6_pantr1')) Below we can see the gene expression for both datatypes. With the interpolated gene expression we’re able to get a better idea as to the cells that are expressing each of the genes. This is especially clear with Pantr1, which clearly localizes to the pyramidal layer. Figure 17.11: Gene expression for visium (left) and interpolated (right) expression for Sparc (top) and Pantr1 (bottom). 17.7.3 Run PCA subcell_v_brain <- runPCA(gobject = subcell_v_brain, spat_unit = "stardist_cell", expression_values = "normalized", feats_to_use = NULL) 17.7.4 Clustering # UMAP subcell_v_brain <- runUMAP(subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:15, n_neighbors = 1000, min_dist = 0.001, spread = 1) # NN Network subcell_v_brain <- createNearestNetwork(gobject = subcell_v_brain, spat_unit = "stardist_cell", dimensions_to_use = 1:10, feats_to_use = hv_feats, expression_values = 'normalized', k = 70) subcell_v_brain <- doLeidenCluster(gobject = subcell_v_brain, spat_unit = "stardist_cell", resolution = 0.15, n_iterations = 100, partition_type = 'RBConfigurationVertexPartition') plotUMAP(subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus') Figure 17.12: UMAP for stardist_cell based on the 1500 interpolated gene expressions. Colored based on leiden clustering. 17.7.5 Visualizing clustering Vizualizing the clustering for both the visium dataset and the interpolated dataset we can similar clusters. However, with the interpolated dataset we are able to see finer detail for each cluster. spatPlot2D(gobject = subcell_v_brain, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.5, point_shape = 'no_border', background_color = 'black', save_plot = F, show_legend = TRUE) spatPlot2D(gobject = subcell_v_brain, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat')) Figure 17.13: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) 17.7.6 Cropping objects We are also able to crop both spat units simultaneosly to zoom in on specific regions of the tissue such as seen below. subcell_v_brain_crop <- subsetGiottoLocs(gobject = subcell_v_brain, spat_unit = ":all:", x_min = 4000, x_max = 7000, y_min = -6500, y_max = -3500, z_max = NULL, z_min = NULL) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'cell', cell_color = 'leiden_clus', show_image = T, point_size = 2, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_vis_spat_crop')) spatPlot2D(gobject = subcell_v_brain_crop, spat_unit = 'stardist_cell', cell_color = 'leiden_clus', show_image = T, point_size = 0.1, point_shape = 'no_border', background_color = 'black', show_legend = TRUE, save_plot = TRUE, save_param = list(save_name = '03_ses6_subcell_spat_crop')) Figure 17.14: Spatial plots showing leiden clustering mapped onto the base visium spots (left) and individual nuceli through interpolation (right) "],["contributing-to-giotto.html", "18 Contributing to Giotto 18.1 Contribution guideline", " 18 Contributing to Giotto Jiaji George Chen August 7th 2024 save_dir <- "~/Documents/GitHub/giotto_workshop_2024/img/03_session7" 18.1 Contribution guideline https://drieslab.github.io/Giotto_website/CONTRIBUTING.html "],["404.html", "Page not found", " Page not found The page you requested cannot be found (perhaps it was moved or renamed). You may want to try searching to find the page's new location, or use the table of contents to find the page you are looking for. "]]
diff --git a/spatial-multi-modal-analysis.html b/spatial-multi-modal-analysis.html
index 5bdab74..c78f952 100644
--- a/spatial-multi-modal-analysis.html
+++ b/spatial-multi-modal-analysis.html
@@ -563,48 +563,48 @@ 13.2.3 Spatial classes:# load in data
-library(Giotto)
-
-g <- GiottoData::loadGiottoMini("vizgen")
-
-activeSpatUnit(g) <- "aggregate"
-
-gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE)
-
-gimg <- getGiottoImage(g)
+
spatPlot2D(visium_tonsil,
- spat_unit = "cell",
- feat_type = "rna",
- cell_color = "integrated_leiden_clus",
- point_size = 3,
- show_image = TRUE,
- title = "Integrated Leiden clustering")spatPlot2D(visium_tonsil,
+ spat_unit = "cell",
+ feat_type = "rna",
+ cell_color = "integrated_leiden_clus",
+ point_size = 3,
+ show_image = TRUE,
+ title = "Integrated Leiden clustering")Figure 14.9: Spatial distribution of the integrated Leiden clusters. @@ -866,85 +866,85 @@
14.8 Multi-omics integration
14.9 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
- [7] matrixStats_1.3.0 compiler_4.4.1
- [9] png_0.1-8 systemfonts_1.1.0
- [11] vctrs_0.6.5 reshape2_1.4.4
- [13] stringr_1.5.1 pkgconfig_2.0.3
- [15] SpatialExperiment_1.14.0 crayon_1.5.3
- [17] fastmap_1.2.0 backports_1.5.0
- [19] magick_2.8.4 XVector_0.44.0
- [21] labeling_0.4.3 utf8_1.2.4
- [23] rmarkdown_2.27 UCSC.utils_1.0.0
- [25] ragg_1.3.2 purrr_1.0.2
- [27] xfun_0.46 beachmat_2.20.0
- [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
- [31] jsonlite_1.8.8 DelayedArray_0.30.1
- [33] BiocParallel_1.38.0 terra_1.7-78
- [35] irlba_2.3.5.1 parallel_4.4.1
- [37] R6_2.5.1 stringi_1.8.4
- [39] RColorBrewer_1.1-3 reticulate_1.38.0
- [41] parallelly_1.37.1 GenomicRanges_1.56.1
- [43] scattermore_1.2 Rcpp_1.0.13
- [45] bookdown_0.40 SummarizedExperiment_1.34.0
- [47] knitr_1.48 future.apply_1.11.2
- [49] R.utils_2.12.3 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
+ [7] matrixStats_1.3.0 compiler_4.4.1
+ [9] png_0.1-8 systemfonts_1.1.0
+ [11] vctrs_0.6.5 reshape2_1.4.4
+ [13] stringr_1.5.1 pkgconfig_2.0.3
+ [15] SpatialExperiment_1.14.0 crayon_1.5.3
+ [17] fastmap_1.2.0 backports_1.5.0
+ [19] magick_2.8.4 XVector_0.44.0
+ [21] labeling_0.4.3 utf8_1.2.4
+ [23] rmarkdown_2.27 UCSC.utils_1.0.0
+ [25] ragg_1.3.2 purrr_1.0.2
+ [27] xfun_0.46 beachmat_2.20.0
+ [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
+ [31] jsonlite_1.8.8 DelayedArray_0.30.1
+ [33] BiocParallel_1.38.0 terra_1.7-78
+ [35] irlba_2.3.5.1 parallel_4.4.1
+ [37] R6_2.5.1 stringi_1.8.4
+ [39] RColorBrewer_1.1-3 reticulate_1.38.0
+ [41] parallelly_1.37.1 GenomicRanges_1.56.1
+ [43] scattermore_1.2 Rcpp_1.0.13
+ [45] bookdown_0.40 SummarizedExperiment_1.34.0
+ [47] knitr_1.48 future.apply_1.11.2
+ [49] R.utils_2.12.3 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
14.9 Session info
- -R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
- [7] matrixStats_1.3.0 compiler_4.4.1
- [9] png_0.1-8 systemfonts_1.1.0
- [11] vctrs_0.6.5 reshape2_1.4.4
- [13] stringr_1.5.1 pkgconfig_2.0.3
- [15] SpatialExperiment_1.14.0 crayon_1.5.3
- [17] fastmap_1.2.0 backports_1.5.0
- [19] magick_2.8.4 XVector_0.44.0
- [21] labeling_0.4.3 utf8_1.2.4
- [23] rmarkdown_2.27 UCSC.utils_1.0.0
- [25] ragg_1.3.2 purrr_1.0.2
- [27] xfun_0.46 beachmat_2.20.0
- [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
- [31] jsonlite_1.8.8 DelayedArray_0.30.1
- [33] BiocParallel_1.38.0 terra_1.7-78
- [35] irlba_2.3.5.1 parallel_4.4.1
- [37] R6_2.5.1 stringi_1.8.4
- [39] RColorBrewer_1.1-3 reticulate_1.38.0
- [41] parallelly_1.37.1 GenomicRanges_1.56.1
- [43] scattermore_1.2 Rcpp_1.0.13
- [45] bookdown_0.40 SummarizedExperiment_1.34.0
- [47] knitr_1.48 future.apply_1.11.2
- [49] R.utils_2.12.3 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7 R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] RcppAnnoy_0.0.22 GiottoUtils_0.1.10
+ [7] matrixStats_1.3.0 compiler_4.4.1
+ [9] png_0.1-8 systemfonts_1.1.0
+ [11] vctrs_0.6.5 reshape2_1.4.4
+ [13] stringr_1.5.1 pkgconfig_2.0.3
+ [15] SpatialExperiment_1.14.0 crayon_1.5.3
+ [17] fastmap_1.2.0 backports_1.5.0
+ [19] magick_2.8.4 XVector_0.44.0
+ [21] labeling_0.4.3 utf8_1.2.4
+ [23] rmarkdown_2.27 UCSC.utils_1.0.0
+ [25] ragg_1.3.2 purrr_1.0.2
+ [27] xfun_0.46 beachmat_2.20.0
+ [29] zlibbioc_1.50.0 GenomeInfoDb_1.40.1
+ [31] jsonlite_1.8.8 DelayedArray_0.30.1
+ [33] BiocParallel_1.38.0 terra_1.7-78
+ [35] irlba_2.3.5.1 parallel_4.4.1
+ [37] R6_2.5.1 stringi_1.8.4
+ [39] RColorBrewer_1.1-3 reticulate_1.38.0
+ [41] parallelly_1.37.1 GenomicRanges_1.56.1
+ [43] scattermore_1.2 Rcpp_1.0.13
+ [45] bookdown_0.40 SummarizedExperiment_1.34.0
+ [47] knitr_1.48 future.apply_1.11.2
+ [49] R.utils_2.12.3 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7 13.2.3 Spatial classes:# load in data
-library(Giotto)
-
-g <- GiottoData::loadGiottoMini("vizgen")
-
-activeSpatUnit(g) <- "aggregate"
-
-gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE)
-
-gimg <- getGiottoImage(g)
# load in data
-library(Giotto)
-
-g <- GiottoData::loadGiottoMini("vizgen")
-
-activeSpatUnit(g) <- "aggregate"
-
-gpoly <- getPolygonInfo(g, return_giottoPolygon = TRUE)
-
-gimg <- getGiottoImage(g)13.3 Examples of the simple transforms with a giottoPolygon
-rain <- rainbow(nrow(gpoly))
-
-line_width <- 0.3
-
-# par to setup the grid plotting layout
-p <- par(no.readonly = TRUE)
-par(mfrow=c(3,3))
-gpoly |>
- plot(main = "no transform", col = rain, lwd = line_width)
-
-gpoly |> spatShift(dx = 1000) |>
- plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
-
-gpoly |> spin(45) |>
- plot(main = "spin(45)", col = rain, lwd = line_width)
-
-gpoly |> rescale(fx = 10, fy = 6) |>
- plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
-
-gpoly |> flip(direction = "vertical") |>
- plot(main = "flip()", col = rain, lwd = line_width)
-
-gpoly |> t() |>
- plot(main = "t()", col = rain, lwd = line_width)
-
-gpoly |> shear(fx = 0.5) |>
- plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
-par(p)rain <- rainbow(nrow(gpoly))
+
+line_width <- 0.3
+
+# par to setup the grid plotting layout
+p <- par(no.readonly = TRUE)
+par(mfrow=c(3,3))
+gpoly |>
+ plot(main = "no transform", col = rain, lwd = line_width)
+
+gpoly |> spatShift(dx = 1000) |>
+ plot(main = "spatShift(dx = 1000)", col = rain, lwd = line_width)
+
+gpoly |> spin(45) |>
+ plot(main = "spin(45)", col = rain, lwd = line_width)
+
+gpoly |> rescale(fx = 10, fy = 6) |>
+ plot(main = "rescale(fx = 10, fy = 6)", col = rain, lwd = line_width)
+
+gpoly |> flip(direction = "vertical") |>
+ plot(main = "flip()", col = rain, lwd = line_width)
+
+gpoly |> t() |>
+ plot(main = "t()", col = rain, lwd = line_width)
+
+gpoly |> shear(fx = 0.5) |>
+ plot(main = "shear(fx = 0.5)", col = rain, lwd = line_width)
+par(p)
13.4 Affine transforms# create affine2d
-aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))
# create affine2d
-aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))# create affine2d
+aff <- affine() # when called without params, this is the same as affine(diag(c(1, 1)))affine2d object also has an anchor spatial extent, which is used in calculations of the translation values. affine2d generates with a default extent, but a specific one matching that of the object you are manipulating (such as that of the giottoPolygon) should be set.# append several simple transforms
-aff <- aff |>
- spatShift(dx = 1000) |>
- spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
- flip(direction = "vertical") |>
- t() |>
- shear(fx = 0.5)
-force(aff)# append several simple transforms
+aff <- aff |>
+ spatShift(dx = 1000) |>
+ spin(45, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ rescale(10, x0 = 0, y0 = 0) |> # without the x0, y0 params, the extent center is used
+ flip(direction = "vertical") |>
+ t() |>
+ shear(fx = 0.5)
+force(aff)<affine2d>
anchor : 6399.24384990901, 6903.24298517207, -5152.38959073896, -4694.86823300896 (xmin, xmax, ymin, ymax)
rotate : -0.785398163397448 (rad)
@@ -639,35 +639,35 @@ 13.4 Affine transformsplot(aff)
We can then apply the affine transforms to the giottoPolygon to see that it indeed in the location and orientation that the projection suggests.
13.5 Image transforms
Giotto uses giottoLargeImages as the core image class which is based on terra SpatRaster. Images are not loaded into memory when the object is generated and instead an amount of regular sampling appropriate to the zoom level requested is performed at time of plotting.
spatShift() and rescale() operations are supported by terra SpatRaster, and we inherit those functionalities. spin(), flip(), t(), shear(), affine() operations will coerce giottoLargeImage to giottoAffineImage, which is much the same, except it contains an affine2d object that tracks spatial manipulations performed, so that they can be applied through magick::image_distort() processing after sampled values are pulled into memory. giottoAffineImage also has alternative ext() and crop() methods so that those operations respect both the expected post-affine space and un-transformed source image.
# affine transform of image info matches with polygon info
-gimg |> affine(aff) |> plot()
-gpoly |> affine(aff) |>
- plot(add = TRUE,
- border = "cyan",
- lwd = 0.3)# affine transform of image info matches with polygon info
+gimg |> affine(aff) |> plot()
+gpoly |> affine(aff) |>
+ plot(add = TRUE,
+ border = "cyan",
+ lwd = 0.3)
# affine of the giotto object
-g |> affine(aff) |>
- spatInSituPlotPoints(
- show_image = TRUE,
- feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
- feats_color_code = rainbow(4),
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- point_size = 0.1,
- use_overlap = FALSE
- )# affine of the giotto object
+g |> affine(aff) |>
+ spatInSituPlotPoints(
+ show_image = TRUE,
+ feats = list(rna = c("Adgrl1", "Gfap", "Ntrk3", "Slc17a7")),
+ feats_color_code = rainbow(4),
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ point_size = 0.1,
+ use_overlap = FALSE
+ )
Currently giotto image objects are not fully compatible with .ome.tif files. terra which relies on gdal drivers for image loading will find that the Gtiff driver opens some .ome.tif images, but fails when certain compressions (notably JP2000 as used by 10x for their single-channel stains) are used.
13.6.1 Example dataset: Xenium Br – cooresponding centroid locations
The goal of creating this multi-modal dataset is to register all the modalities listed above to the same coordinate system as Xenium in situ transcripts as the coordinate represents a certain micron distance.
-library(Giotto)
-instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
- save_plot = TRUE,
- show_plot = TRUE)
-options(timeout = 999999)
-download_dir <-file.path(getwd(),'/data/03_session2/')
-destfile <- file.path(download_dir,'Multimodal_registration.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
-Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
-Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')library(Giotto)
+instrs <- createGiottoInstructions(save_dir = file.path(getwd(),'/img/03_session2/'),
+ save_plot = TRUE,
+ show_plot = TRUE)
+options(timeout = 999999)
+download_dir <-file.path(getwd(),'/data/03_session2/')
+destfile <- file.path(download_dir,'Multimodal_registration.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13208139/files/Multimodal_registration.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Multimodal_registration.zip'), exdir = download_dir)
+Xenium_dir <- paste0(download_dir,'/Multimodal_registration/Xenium/')
+Visium_dir <- paste0(download_dir,'/Multimodal_registration/Visium/')13.6.2 Target Coordinate system
Xenium transcripts, polygon information and corresponding centroids are output from the Xenium instrument and are in the same coordinate system from the raw output. We can start with checking the centroid information as a representation of the target coordinate system.
-xen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
-xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
-xen_cell_plxen_cell_df <- read.csv(paste0(Xenium_dir,"cells.csv.gz"))
+xen_cell_pl <- ggplot2::ggplot() + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
+xen_cell_pl
13.6.3 Visium to register
Load the Visium directory using the Giotto convenience function, note that here we are using the “tissue_hires_image.png” as a image to plot. Using the convenience function, hires image and scale factor stored in the spaceranger will be used for automatic alignment while the Giotto Visium Object creation. SpatPlot2D provided by Giotto will random sample pixels from the image you provide, thus providing microscopic image as the image input for createGiottoVisiumObject() will improve the visual performance for downstream registration
-G_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
- gene_column_index = 2,
- png_name = 'tissue_hires_image.png',
- instructions = NULL)
-# In the meantime, calculate statistics for easier plot showing
-G_visium <- normalizeGiotto(G_visium)
-G_visium <- addStatistics(G_visium)
-V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
-V_originG_visium <- createGiottoVisiumObject(visium_dir = Visium_dir,
+ gene_column_index = 2,
+ png_name = 'tissue_hires_image.png',
+ instructions = NULL)
+# In the meantime, calculate statistics for easier plot showing
+G_visium <- normalizeGiotto(G_visium)
+G_visium <- addStatistics(G_visium)
+V_origin <- spatPlot2D(G_visium,show_image = T,point_size = 0,return_plot = T)
+V_origin
The Visium Object needs to be transformed to the same orientation as target coordinate system, so we perform the first transform.
-# create affine2d
-aff <- affine(diag(c(1,1)))
-aff <- aff |>
- spin(90) |>
- flip(direction = "horizontal")
-force(aff)
-
-# Apply the transform
-V_tansformed <- affine(G_visium,aff)
-spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
-spatplot_to_register# create affine2d
+aff <- affine(diag(c(1,1)))
+aff <- aff |>
+ spin(90) |>
+ flip(direction = "horizontal")
+force(aff)
+
+# Apply the transform
+V_tansformed <- affine(G_visium,aff)
+spatplot_to_register <- spatPlot2D(V_tansformed,show_image = T,point_size = 0,return_plot = T)
+spatplot_to_register
Landmarks are considered to be a set of points that are defining same location from two different resources. They are very helpful to be used as anchors to create affine transformtion. For example, after the affine transformation source landmarks should be as close to target landmarks as possible. Since images from different modalities can share similar morphology, the easiest way is to pin landmarks at the morphological identities shared between images. Giotto provides a interactive landmark selection tool to pin landmarks, two input plots can be generated from a ggplot object, a GiottoLargeImage object, or a path to a image you want to register for. Note that if you directly provide image path, you will need to create a separate GiottoLargeImage to perform transformation, and make sure the GiottoLargeImage has the same coordinate system as shown in the shiny app.
- +
Now, use the landmarks to estimate the transformation matrix needed, and to register the Giotto Visium Object to the target coordinate system. For reproducibility purpose, the landmarks used in the chunck below will be loaded from saved result.
-landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
-affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
-
-V_final <- affine(G_visium,affine_mtx %*% aff@affine)
-
-spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
-spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()landmarks<- readRDS(paste0(Xenium_dir,'/Visium_to_Xen_Landmarks.rds'))
+affine_mtx <- calculateAffineMatrixFromLandmarks(landmarks[[1]],landmarks[[2]])
+
+V_final <- affine(G_visium,affine_mtx %*% aff@affine)
+
+spatplot_final <- spatPlot2D(V_final,show_image = T,point_size = 0,show_plot = F)
+spatplot_final + ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.3.1 Create Pseudo Visium dataset for comparison
Giotto provides a way to create different shapes on certain locations, we can use that to create a pseudo-visium polygons to aggregate transcripts or image intensities. To do that, we will need the centroid locations, which can be get using getSpatialLocations(). and also the radius information to create circles. We know that Visium certer to center distance is 100um and spot diameter is 55um, thus we can estimate the radius from certer to center distance. And we can use a spatial network created by nearest neighbor = 2 to capture the distance.
-V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
-spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
-spatPlot2D(V_final,
- show_network = T,
- network_color = 'blue',
- point_size = 1)
-center_to_center <- min(spat_network$distance)
-radius <- center_to_center*55/200V_final <- createSpatialNetwork(V_final, k = 1,method = 'kNN')
+spat_network <- getSpatialNetwork(V_final,output = 'networkDT')
+spatPlot2D(V_final,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1)
+center_to_center <- min(spat_network$distance)
+radius <- center_to_center*55/200
Now we get the Pseudo Visium polygons
-Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
-stamp_dt <- circleVertices(radius = radius, npoints = 100)
-pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
-pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
-plot(pseudo_visium_poly)Visium_centroid <- getSpatialLocations(V_final,output = 'data.table')
+stamp_dt <- circleVertices(radius = radius, npoints = 100)
+pseudo_visium_dt <- polyStamp(stamp_dt, Visium_centroid)
+pseudo_visium_poly <- createGiottoPolygonsFromDfr(pseudo_visium_dt,calc_centroids = T)
+plot(pseudo_visium_poly)
Create Xenium object with pseudo Visium polygon. To save run time, example shown here only have MS4A1 and ERBB2 genes to create Giotto points
-xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
-gpoints <- createGiottoPoints(xen_transcripts)
-Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
- gpolygons = list('visium' = pseudo_visium_poly))xen_transcripts <- data.table::fread(paste0(Xenium_dir,'Xen_2_genes.csv.gz'))
+gpoints <- createGiottoPoints(xen_transcripts)
+Xen_obj <-createGiottoObjectSubcellular(gpoints = list('rna' = gpoints),
+ gpolygons = list('visium' = pseudo_visium_poly))Get gene expression information by overlapping polygon to points
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'visium')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "visium",
- feat_info = "rna",
- aggr_function = "sum")Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'visium')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "visium",
+ feat_info = "rna",
+ aggr_function = "sum")Manipulate the expression for plotting
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'visium',
- name = 'plot')
-
-spatFeatPlot2D(Xen_obj,
- point_size = 3.5,
- expression_values = 'plot',
- show_image = F,
- feats = 'ERBB2')Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'visium',
+ name = 'plot')
+
+spatFeatPlot2D(Xen_obj,
+ point_size = 3.5,
+ expression_values = 'plot',
+ show_image = F,
+ feats = 'ERBB2')
Subset the registered Visium and plot same gene
-#get the extent of giotto points, xmin, xmax, ymin, ymax
-subset_extent <- ext(gpoints@spatVector)
-sub_visium <- subsetGiottoLocs(V_final,
- x_min = subset_extent[1],
- x_max = subset_extent[2],
- y_min = subset_extent[3],
- y_max = subset_extent[4])
-spatFeatPlot2D(sub_visium,
- point_size = 2,
- expression_values = 'scaled',
- show_image = F,
- feats = 'ERBB2')#get the extent of giotto points, xmin, xmax, ymin, ymax
+subset_extent <- ext(gpoints@spatVector)
+sub_visium <- subsetGiottoLocs(V_final,
+ x_min = subset_extent[1],
+ x_max = subset_extent[2],
+ y_min = subset_extent[3],
+ y_max = subset_extent[4])
+spatFeatPlot2D(sub_visium,
+ point_size = 2,
+ expression_values = 'scaled',
+ show_image = F,
+ feats = 'ERBB2')
13.6.4 Register post-Xenium H&E and IF image
For Xenium instrument output, Giotto provide a convenience function to load the output from the Xenium ranger output. Note that 10X created the affine image alignment file by applying rotation, scale at (0,0) of the top left corner and translation last. Thus, it will look different than the affine matrix created from landmarks above. In this example, we used a 0.05X compressed ometiff and the alignment file is also create by first rescale at 20X, then apply the affine matrix provided by 10X Genomics.
-HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
- imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
- micron = 0.2125)
-plot(HE_xen)HE_xen <- read10xAffineImage(file = paste0(Xenium_dir, "HE_ome_compressed.tiff"),
+ imagealignment_path = paste0(Xenium_dir,"Xenium_he_imagealignment.csv"),
+ micron = 0.2125)
+plot(HE_xen)
The image is still on the top left corner, so we flip the image to make it align with the target coordinate system. We can also save the transformed image raster by re-sample all pixel from the original image, and write it to a file on disk for future use.
-HE_xen <- HE_xen |> flip(direction = "vertical")
-gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
-plot(gimg_rast)
-#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)HE_xen <- HE_xen |> flip(direction = "vertical")
+gimg_rast <- HE_xen@funs$realize_magick(size = prod(dim(HE_xen)))
+plot(gimg_rast)
+#terra::writeRaster(gimg_rast@raster_object, filename = output, gdal = "COG" # save as GeoTIFF with extent info)
Now we can check the registration results. GiottoVisuals provide a function to plot a giottoLargeImage to a ggplot object in order to plot additional layers of ggplots
-gg <- ggplot2::ggplot()
-pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
-pl + ggplot2::geom_smooth() +
- ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()gg <- ggplot2::ggplot()
+pl <- GiottoVisuals::gg_annotation_raster(gg,gimg_rast)
+pl + ggplot2::geom_smooth() +
+ ggplot2::geom_point(data = xen_cell_df, ggplot2::aes(x = x_centroid , y = y_centroid),size = 1e-150,,color = 'orange') + ggplot2::theme_classic()
13.6.4.1 Add registered image information and compare RNA vs protein expression
With the strategy described above, affine transformed image can be saved and used for quantitive analysis. Here, we can use the same strategy as dealing with spatial proteomics data for IF
-CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
-HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
-Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
- largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))CD20_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/CD20_registered.tiff'), use_rast_ext = T,name = 'CD20')
+HER2_gimg <- createGiottoLargeImage(paste0(Xenium_dir,'/HER2_registered.tiff'), use_rast_ext = T,name = 'HER2')
+Xen_obj <- addGiottoLargeImage(gobject = Xen_obj,
+ largeImages = list('CD20' = CD20_gimg,'HER2' = HER2_gimg))Get the cell polygons, as Xenium and IF are both subcellular resolution
-cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
-colnames(cellpoly_dt) <- c('poly_ID','x','y')
-cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
-Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))cellpoly_dt <- data.table::fread(paste0(Xenium_dir,'cell_boundaries.csv.gz'))
+colnames(cellpoly_dt) <- c('poly_ID','x','y')
+cellpoly <- createGiottoPolygonsFromDfr(cellpoly_dt)
+Xen_obj <- addGiottoPolygons(Xen_obj,gpolygons = list('cell' = cellpoly))Compute the gene expression matrix by overlay the cell polygons and giotto points.
-Xen_obj <- calculateOverlap(Xen_obj,
- feat_info = 'rna',
- spatial_info = 'cell')
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "point",
- poly_info = "cell",
- feat_info = "rna",
- aggr_function = "sum")
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'rna',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'rna',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'rna',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'ERBB2',
- point_size = 0.05)Xen_obj <- calculateOverlap(Xen_obj,
+ feat_info = 'rna',
+ spatial_info = 'cell')
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "point",
+ poly_info = "cell",
+ feat_info = "rna",
+ aggr_function = "sum")
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'rna',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'rna',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'ERBB2',
+ point_size = 0.05)
Now we overlay the HER2 expression from the raster image with the cell polygons.
-Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'cell',
- image_names = c('HER2','CD20'))
-
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'cell',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'cell',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'cell',
- feats = 'HER2',
- point_size = 0.05)Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'cell',
+ image_names = c('HER2','CD20'))
+
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'cell',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'cell',
+ feats = 'HER2',
+ point_size = 0.05)
We can also overlay the protein expression to Visium spots
Xen_obj <- calculateOverlap(Xen_obj,
- spatial_info = 'visium',
- image_names = c('HER2','CD20'))
-Xen_obj <- overlapToMatrix(x = Xen_obj,
- type = "intensity",
- poly_info = "visium",
- feat_info = "protein",
- aggr_function = "sum")
-
-Xen_obj <- filterGiotto(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- expression_threshold = 1,
- feat_det_in_min_cells = 0,
- min_det_feats_per_cell = 1)
-
-tmp_exprs <- getExpression(Xen_obj,
- feat_type = 'protein',
- spat_unit = 'visium',
- output = 'matrix')
-Xen_obj <- setExpression(Xen_obj,
- x = createExprObj(log(tmp_exprs+1)),
- feat_type = 'protein',
- spat_unit = 'visium',
- name = 'plot')
-spatFeatPlot2D(Xen_obj,
- feat_type = 'protein',
- expression_values = 'plot',
- spat_unit = 'visium',
- feats = 'HER2',
- point_size = 2)Xen_obj <- calculateOverlap(Xen_obj,
+ spatial_info = 'visium',
+ image_names = c('HER2','CD20'))
+Xen_obj <- overlapToMatrix(x = Xen_obj,
+ type = "intensity",
+ poly_info = "visium",
+ feat_info = "protein",
+ aggr_function = "sum")
+
+Xen_obj <- filterGiotto(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ expression_threshold = 1,
+ feat_det_in_min_cells = 0,
+ min_det_feats_per_cell = 1)
+
+tmp_exprs <- getExpression(Xen_obj,
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ output = 'matrix')
+Xen_obj <- setExpression(Xen_obj,
+ x = createExprObj(log(tmp_exprs+1)),
+ feat_type = 'protein',
+ spat_unit = 'visium',
+ name = 'plot')
+spatFeatPlot2D(Xen_obj,
+ feat_type = 'protein',
+ expression_values = 'plot',
+ spat_unit = 'visium',
+ feats = 'HER2',
+ point_size = 2)
13.6.4.1 Add registered image inf
13.6.5 Automatic alignment via SIFT feature descriptor matching and affine transformation
Pin landmarks or use compounded affine transforms to register image usually provides initial registration results. However, recording landmarks or manually combine transformations require a lot of manual effort. It will require too much effort when having a large amount of images to register. As long as accurate landmarks are provided, registration will be easy to automatically perform. Here we provide a wrapper function of Scale invariant feature transform(SIFT). SIFT will first identify the extreme points in different scale spaces from paired images, then use a brutal force way to match the points. The matched points can then be used to estimate the transform and warp the image. The major drawback is once the dimension of the image become bigger, the computing time will increase exponentially.
Here, we provide an example of two compressed images to show the automatic alignment pipeline.
- +
IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
-terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")IF <- createGiottoLargeImage(paste0(Xenium_dir,'mini_IF.tif'),negative_y = F,flip_horizontal = T)
+terra::plotRGB(IF@raster_object,r=1, g=2, b=3,, stretch="lin")
Now, we can use the automated transformation pipeline. Note that we will start with a path to the images, run the preprocessImageToMatrix() first to meet the requirement of estimateAutomatedImageRegistrationWithSIFT() function. The images will be preprocessed to gray scale. And for that purpose, we use the DAPI channel from the miniIF, and set invert = T for mini HE as HE image so that grayscle HE will have higher value for high intensity pixels. The function will output an estimation of the transform.
-estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
- flip_horizontal = T,
- use_single_channel = T,
- single_channel_number = 3),
- y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
- invert = T),
- plot_match = T,
- max_ratio = 0.5,estimate_fun = 'Projective')estimation <- estimateAutomatedImageRegistrationWithSIFT(x = preprocessImageToMatrix(paste0(Xenium_dir,'mini_IF.tif'),
+ flip_horizontal = T,
+ use_single_channel = T,
+ single_channel_number = 3),
+ y = preprocessImageToMatrix(paste0(Xenium_dir,'mini_HE.png'),
+ invert = T),
+ plot_match = T,
+ max_ratio = 0.5,estimate_fun = 'Projective')
Use the estimation, we can quickly visualize the transformation
-mtx <- as.matrix(estimation$params)
-transformed <- affine(IF, mtx)
-To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
-plot(HE)
-plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
-SIFT_registration_overlaymtx <- as.matrix(estimation$params)
+transformed <- affine(IF, mtx)
+To_see_overlay <- transformed@funs$realize_magick(size = 2e6)
+plot(HE)
+plot(To_see_overlay@raster_object[[2]], add=TRUE, alpha=0.5)
+SIFT_registration_overlay
diff --git a/spatial-proteomics-multiplexed-immunofluorescence.html b/spatial-proteomics-multiplexed-immunofluorescence.html
index 684d4e8..f550962 100644
--- a/spatial-proteomics-multiplexed-immunofluorescence.html
+++ b/spatial-proteomics-multiplexed-immunofluorescence.html
@@ -518,33 +518,33 @@
+
+
11.2 Raw data type coming out of
@@ -618,34 +618,34 @@ 11.3 Cell Segmentation file to ge
@@ -654,133 +654,133 @@ 11 Spatial proteomics: Multiplexe
Junxiang Xu
August 6th 2024
Before you start, this tutorial contains an optional part to run image segmentation using Giotto wrapper of Cellpose. If considering to use that function, please restart R session as we will need to activate a new Giotto python environment. The environment is also compatible with other Giotto functions. We will also need to install the Cellpose supported Giotto environment if haven’t done so.
-#Install the Giotto Environment with Cellpose, note that we only need to do it once
-reticulate::conda_create(envname = "giotto_cellpose",
- python_version = 3.8)
-#.re.restartR()
-reticulate::use_condaenv('giotto_cellpose')
-reticulate::py_install(
- pip = TRUE,
- envname = 'giotto_cellpose',
- packages = c(
- "pandas",
- "networkx",
- "python-igraph",
- "leidenalg",
- "scikit-learn",
- "cellpose",
- "smfishhmrf",
- 'tifffile',
- 'scikit-image'
- )
-)
-#.rs.restartR()#Install the Giotto Environment with Cellpose, note that we only need to do it once
+reticulate::conda_create(envname = "giotto_cellpose",
+ python_version = 3.8)
+#.re.restartR()
+reticulate::use_condaenv('giotto_cellpose')
+reticulate::py_install(
+ pip = TRUE,
+ envname = 'giotto_cellpose',
+ packages = c(
+ "pandas",
+ "networkx",
+ "python-igraph",
+ "leidenalg",
+ "scikit-learn",
+ "cellpose",
+ "smfishhmrf",
+ 'tifffile',
+ 'scikit-image'
+ )
+)
+#.rs.restartR()Now, activate the Giotto python environment.
-#.rs.restartR()
-# Activate the Giotto python environment of your choice
-GiottoClass::set_giotto_python_path('giotto_cellpose')
-# Check if cellpose was successfully installed
-GiottoUtils::package_check('cellpose',repository = 'pip')#.rs.restartR()
+# Activate the Giotto python environment of your choice
+GiottoClass::set_giotto_python_path('giotto_cellpose')
+# Check if cellpose was successfully installed
+GiottoUtils::package_check('cellpose',repository = 'pip')11.1 Spatial Proteomics Technologies
This tutorial is aimed at analyzing spatially resolved multiplexed immunofluorescence data. It is compatible for different kinds of image based spatial proteomics data, such as Akoya(CODEX), CyCIF, IMC, MIBI, and Lunaphore(seqIF). Note that this tutorial will focus on starting directly with the intensity data(image), not the decoded count matrix. @@ -557,58 +557,58 @@
11.2 Raw data type coming out of
11.2.1 Use ome.tiff as an example output data to begin with
OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore.
-library(Giotto)
-instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
- save_plot = TRUE,
- show_plot = TRUE,
- python_path = 'giotto_cellpose')
-options(timeout = 9999999)
-download_dir =file.path(getwd(),'/data/02_session4/')
-destfile = file.path(download_dir,'Lunaphore.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
-list.files(paste0(download_dir,'/Lunaphore'))
+library(Giotto)
+instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ python_path = 'giotto_cellpose')
+options(timeout = 9999999)
+download_dir =file.path(getwd(),'/data/02_session4/')
+destfile = file.path(download_dir,'Lunaphore.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
+list.files(paste0(download_dir,'/Lunaphore'))
We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML.
-img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
-img_meta
+img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
+img_meta
However, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it.
-##Get channel information
-Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
-node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
-channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
-channel_df
+##Get channel information
+Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
+node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
+channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
+channel_df
11.2.2 Use single channel images as an example output data to begin with
Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like.
-#Create multichannel raster and extract each single channels
-Luna_terra = terra::rast(Luna)
-names(Luna_terra) <- channel_df$Name
-gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
-plot(gimg_DAPI)
+#Create multichannel raster and extract each single channels
+Luna_terra = terra::rast(Luna)
+names(Luna_terra) <- channel_df$Name
+gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
+plot(gimg_DAPI)
Extract and save the raster image for future use.
-single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
-if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
-for (i in 1:nrow(channel_df)){
- single_channel <- terra::subset(Luna_terra, i)
- terra::writeRaster(single_channel,
- filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
- overwrite=T)
-}
+single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
+if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
+for (i in 1:nrow(channel_df)){
+ single_channel <- terra::subset(Luna_terra, i)
+ terra::writeRaster(single_channel,
+ filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
+ overwrite=T)
+}
Create a list of GiottoLargeImages using single channel rasters.
-file_names = list.files(single_channel_dir,full.names = TRUE)
-image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
-gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
- names = image_names,
- negative_y = F,
- flip_vertical = T)
-names(gimg_list) <- image_names
-plot(gimg_list[['Vimentin']])
+file_names = list.files(single_channel_dir,full.names = TRUE)
+image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
+gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
+ names = image_names,
+ negative_y = F,
+ flip_vertical = T)
+names(gimg_list) <- image_names
+plot(gimg_list[['Vimentin']])
11.2.1 Use ome.tiff as an example output data to begin with
OME-TIFF (Open Microscopy Environment Tagged Image File Format) is a file format designed for including detailed metadata and support for multi-dimensional image data. This is a common output file format for spatial proteomics platform such as lunaphore.
-library(Giotto)
-instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
- save_plot = TRUE,
- show_plot = TRUE,
- python_path = 'giotto_cellpose')
-options(timeout = 9999999)
-download_dir =file.path(getwd(),'/data/02_session4/')
-destfile = file.path(download_dir,'Lunaphore.zip')
-if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
-download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
-unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
-list.files(paste0(download_dir,'/Lunaphore'))library(Giotto)
+instrs = createGiottoInstructions(save_dir = file.path(getwd(),'/img/02_session4/'),
+ save_plot = TRUE,
+ show_plot = TRUE,
+ python_path = 'giotto_cellpose')
+options(timeout = 9999999)
+download_dir =file.path(getwd(),'/data/02_session4/')
+destfile = file.path(download_dir,'Lunaphore.zip')
+if (!dir.exists(download_dir)) { dir.create(download_dir, recursive = TRUE) }
+download.file('https://zenodo.org/records/13175721/files/Lunaphore.zip?download=1', destfile = destfile)
+unzip(paste0(download_dir,'/Lunaphore.zip'), exdir = download_dir)
+list.files(paste0(download_dir,'/Lunaphore'))We provide a way to extract meta data information directly from ome.tiffs. Please note that different platforms may store the meta data such as channel information in a different format, we will probably need to change the node names of the ome-XML.
-img_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
-img_metaimg_path <- paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+img_meta <- ometif_metadata(img_path, node = "Channel", output = "data.frame")
+img_metaHowever, sometimes a cropping could result in a loss of ome-XML information from the ome.tiff file. That way, we can use a different strategy to parse the xml information seperately and get channel information from it.
-##Get channel information
-Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
-xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
-node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
-channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
-channel_df##Get channel information
+Luna = paste0(download_dir,"Lunaphore/Lunaphore_example.ome.tiff")
+xmldata = xml2::read_xml(paste0(download_dir,"Lunaphore/Lunaphore_sample_metadata.xml"))
+node <- xml2::xml_find_all(xmldata, "//d1:Channel", ns = xml2::xml_ns(xmldata))
+channel_df <- as.data.frame(Reduce("rbind", xml2::xml_attrs(node)))
+channel_df
11.2.2 Use single channel images as an example output data to begin with
Some platforms may also deconvolute and output gray scale single channel images. And we can create single channel images from ome.tiffs, the single channel images will be of the same format if the platform provide single channel gray scale images. With the single channel images, we can create a GiottoLargeImage and see what it looks like.
-#Create multichannel raster and extract each single channels
-Luna_terra = terra::rast(Luna)
-names(Luna_terra) <- channel_df$Name
-gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
-plot(gimg_DAPI)#Create multichannel raster and extract each single channels
+Luna_terra = terra::rast(Luna)
+names(Luna_terra) <- channel_df$Name
+gimg_DAPI = createGiottoLargeImage(Luna_terra[[1]],negative_y = F,flip_vertical = T)
+plot(gimg_DAPI)
Extract and save the raster image for future use.
-single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
-if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
-for (i in 1:nrow(channel_df)){
- single_channel <- terra::subset(Luna_terra, i)
- terra::writeRaster(single_channel,
- filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
- overwrite=T)
-}single_channel_dir = paste0(download_dir,'/Lunaphore/single_channels/')
+if (!dir.exists(single_channel_dir)) { dir.create(single_channel_dir, recursive = TRUE) }
+for (i in 1:nrow(channel_df)){
+ single_channel <- terra::subset(Luna_terra, i)
+ terra::writeRaster(single_channel,
+ filename = paste0(single_channel_dir,names(single_channel),'.tiff'),
+ overwrite=T)
+}Create a list of GiottoLargeImages using single channel rasters.
-file_names = list.files(single_channel_dir,full.names = TRUE)
-image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
-gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
- names = image_names,
- negative_y = F,
- flip_vertical = T)
-names(gimg_list) <- image_names
-plot(gimg_list[['Vimentin']])file_names = list.files(single_channel_dir,full.names = TRUE)
+image_names = sub("\\.tiff$", "", list.files(single_channel_dir))
+gimg_list <- createGiottoLargeImageList(raster_objects = file_names,
+ names = image_names,
+ negative_y = F,
+ flip_vertical = T)
+names(gimg_list) <- image_names
+plot(gimg_list[['Vimentin']])
11.3 Cell Segmentation file to ge
11.3.1 Using segmentation output file from DeepCell(mesmer) as an example.
We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function.
-gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
-plot(gpoly_mesmer)
+gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
+plot(gpoly_mesmer)
We can also zoom in to check how does the segmentation look.
-zoom <- c(2000,2500,2000,2500)
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+zoom <- c(2000,2500,2000,2500)
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
11.3.2 Using Giotto wrapper of Cellpose to perform segmentation
-gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
-writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
-#Create a giotto image to evaluate segmentation
-gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
+gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
+writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
+#Create a giotto image to evaluate segmentation
+gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)
Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website
-doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
- mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- channel_1 = 0,
- channel_2 = 0,
- model_name = 'cyto3',
- batch_size=12)
-cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- shift_horizontal_step = F,
- shift_vertical_step = F,
- flip_vertical = T)
-plot(gimg_for_cellpose)
-plot(cpoly, add = T, border = 'red')
+doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
+ mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ channel_1 = 0,
+ channel_2 = 0,
+ model_name = 'cyto3',
+ batch_size=12)
+cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ shift_horizontal_step = F,
+ shift_vertical_step = F,
+ flip_vertical = T)
+plot(gimg_for_cellpose)
+plot(cpoly, add = T, border = 'red')
11.3.1 Using segmentation output file from DeepCell(mesmer) as an example.
We collapsed several different channels to created a pseudo memberane staining channel(“nuc_and_bound.tif” provided here), and use that as an input for the deepcell mesmer segmentation pipeline. We can load the output mask from to GiottoPolygon via a convenience function.
-gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
-plot(gpoly_mesmer)gpoly_mesmer = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/whole_cell_mask.tif'),shift_horizontal_step = F,shift_vertical_step = F,flip_vertical = T,calc_centroids = T)
+plot(gpoly_mesmer)
We can also zoom in to check how does the segmentation look.
-zoom <- c(2000,2500,2000,2500)
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)zoom <- c(2000,2500,2000,2500)
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
11.3.2 Using Giotto wrapper of Cellpose to perform segmentation
-gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
-writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
-#Create a giotto image to evaluate segmentation
-gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)gimg_cropped <- cropGiottoLargeImage(giottoLargeImage = gimg_DAPI, crop_extent = terra::ext(zoom))
+writeGiottoLargeImage(gimg_cropped, filename = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),overwrite = T)
+#Create a giotto image to evaluate segmentation
+gimg_for_cellpose <- createGiottoLargeImage(paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),negative_y = F)Now we can run the cellpose segmentation. We can provide different parameters for cellpose inference model(flow_threshold,cellprob_threshold,etc), and practically, the batch size represents how many 224X224 images are calculated in parallel, increasing the amount will increase RAM/VRAM reqirement, lowering the amount will increase the run time.For more information please refer to the cellpose website
-doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
- mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- channel_1 = 0,
- channel_2 = 0,
- model_name = 'cyto3',
- batch_size=12)
-cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
- shift_horizontal_step = F,
- shift_vertical_step = F,
- flip_vertical = T)
-plot(gimg_for_cellpose)
-plot(cpoly, add = T, border = 'red')doCellposeSegmentation(image_dir = paste0(download_dir,'/Lunaphore/DAPI_forcellpose.tiff'),
+ mask_output = paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ channel_1 = 0,
+ channel_2 = 0,
+ model_name = 'cyto3',
+ batch_size=12)
+cpoly = createGiottoPolygonsFromMask(paste0(download_dir,'/Lunaphore/giotto_cellpose_seg.tiff'),
+ shift_horizontal_step = F,
+ shift_vertical_step = F,
+ flip_vertical = T)
+plot(gimg_for_cellpose)
+plot(cpoly, add = T, border = 'red')
11.4 Create a Giotto Object using
You will need to have - list of giotto images - giottoPolygon created from segmentation
-Lunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
- images = gimg_list,
- instructions = instrs)
-Lunaphore_giottoLunaphore_giotto = createGiottoObjectSubcellular(gpolygons = list('cell' = gpoly_mesmer),
+ images = gimg_list,
+ instructions = instrs)
+Lunaphore_giotto11.4.1 Overlap to matrix
calculateOverlap() and overlapToMatrix() are used to overlap the intensity values with
-Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
- spatial_info = 'cell',
- image_names = names(gimg_list))
-
-Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
- type ='intensity',
- poly_info = "cell",
- feat_info = "protein",
- aggr_function = "sum")
-showGiottoExpression(Lunaphore_giotto)Lunaphore_giotto = calculateOverlap(Lunaphore_giotto,
+ spatial_info = 'cell',
+ image_names = names(gimg_list))
+
+Lunaphore_giotto = overlapToMatrix(x = Lunaphore_giotto,
+ type ='intensity',
+ poly_info = "cell",
+ feat_info = "protein",
+ aggr_function = "sum")
+showGiottoExpression(Lunaphore_giotto)11.4.2 Manipulate Expression information
For IF data, DAPI staining is usally only used for stain nuclei, the intensity value of DAPI usually does not have meaningful result to drive difference between cell types. Similar things could happen when a platform uses some reference channel to adjust signal calling, such as TRITC or Cy5, These images will be loaded but need to be removed for expression profile.
Therefore, we could extract the feature expression matrix, filter the DAPI information and write it back to the Giotto Object
expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
-filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
-Lunaphore_giotto <- setExpression(Lunaphore_giotto,
- feat_type = 'protein',
- x = createExprObj(filtered_expr_mtx),
- name = 'raw')
-showGiottoExpression(Lunaphore_giotto)expr_mtx <- getExpression(Lunaphore_giotto, values = 'raw', output = 'matrix')
+filtered_expr_mtx <- expr_mtx[rownames(expr_mtx) != 'DAPI',]
+Lunaphore_giotto <- setExpression(Lunaphore_giotto,
+ feat_type = 'protein',
+ x = createExprObj(filtered_expr_mtx),
+ name = 'raw')
+showGiottoExpression(Lunaphore_giotto)11.4.3 Rescale polygons
rescalePolygons() will provide a quick way to manipulate the polygon size and potentially affect the expression for each cell. redo the calculateOverlap() and overlapToMatrix() will potentially change the downstream analysis
-Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
- poly_info = 'cell',
- name = 'smallcell',
- fx = 0.7, fy = 0.7, calculate_centroids = T)
-smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
-plot(gimg_DAPI,ext = zoom)
-plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
-plot(smallpoly,add = TRUE, border = "red",ext = zoom)Lunaphore_giotto = rescalePolygons(gobject = Lunaphore_giotto,
+ poly_info = 'cell',
+ name = 'smallcell',
+ fx = 0.7, fy = 0.7, calculate_centroids = T)
+smallpoly = get_polygon_info(Lunaphore_giotto,polygon_name = 'smallcell')
+plot(gimg_DAPI,ext = zoom)
+plot(gpoly_mesmer,add = TRUE, border = "white",ext = zoom)
+plot(smallpoly,add = TRUE, border = "red",ext = zoom)
11.4.4 Perform clustering and differential expression
The Giotto Object can then go through standard analysis pipeline normalization, dimensional reduction and clustering
-Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
-Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
-
-Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
- scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
- set_seed = TRUE)
-screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)Lunaphore_giotto <- normalizeGiotto(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein')
+Lunaphore_giotto = addStatistics(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein')
+
+Lunaphore_giotto = runPCA(Lunaphore_giotto, spat_unit = 'cell', feat_type = 'protein',
+ scale_unit = F, center = F, ncp = 20,feats_to_use = NULL,
+ set_seed = TRUE)
+screePlot(Lunaphore_giotto,spat_unit = 'cell', feat_type = 'protein',show_plot = T)
Due to the limited number of total features we have, Leiden clustering generally does not work very well compared to Kmeans or hirechical clustering. Here we can use hirechical clustering to do a quick check.
-Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
- dimensions_to_use = 1:5,
- set_seed = TRUE)
-Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
-Lunaphore_giotto = doHclust(Lunaphore_giotto,
- k = 8,
- dim_reduction_to_use = 'cells',
- spat_unit = 'cell',
- feat_type = 'protein')
-
-spatInSituPlotPoints(gobject = Lunaphore_giotto,
- spat_unit = "cell",
- polygon_feat_type = "cell",
- show_polygon = T,
- feat_type = "protein",
- feats = NULL,
- polygon_fill = 'hclust',
- polygon_fill_as_factor = TRUE,
- polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
- polygon_color = 'black',
- background_color = "white")Lunaphore_giotto = runUMAP(gobject = Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein',
+ dimensions_to_use = 1:5,
+ set_seed = TRUE)
+Lunaphore_giotto = createNearestNetwork(Lunaphore_giotto, spat_unit = 'cell',feat_type = 'protein', dimensions_to_use = 1:5)
+Lunaphore_giotto = doHclust(Lunaphore_giotto,
+ k = 8,
+ dim_reduction_to_use = 'cells',
+ spat_unit = 'cell',
+ feat_type = 'protein')
+
+spatInSituPlotPoints(gobject = Lunaphore_giotto,
+ spat_unit = "cell",
+ polygon_feat_type = "cell",
+ show_polygon = T,
+ feat_type = "protein",
+ feats = NULL,
+ polygon_fill = 'hclust',
+ polygon_fill_as_factor = TRUE,
+ polygon_line_size = 0,largeImage_name = c('CD68'),show_image = T,return_plot = T,
+ polygon_color = 'black',
+ background_color = "white")
Then we can check the heatmap of protein expression and determine the first round of cluster annotation
-cluster_column = 'hclust'
-plotMetaDataHeatmap(Lunaphore_giotto,
- spat_unit = 'cell',feat_type = 'protein',
- expression_values = "raw",
- metadata_cols = c(cluster_column),
- selected_feats = names(gimg_list),
- y_text_size = 8,
- show_values = 'zscores_rescaled')cluster_column = 'hclust'
+plotMetaDataHeatmap(Lunaphore_giotto,
+ spat_unit = 'cell',feat_type = 'protein',
+ expression_values = "raw",
+ metadata_cols = c(cluster_column),
+ selected_feats = names(gimg_list),
+ y_text_size = 8,
+ show_values = 'zscores_rescaled')
11.4.5 Give the cluster an annotation based on expression values
- +11.4.6 spatial network
This is to create a cellular neighborhood based on nearest neighbor of physical distance
-Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
-
-spatPlot2D(Lunaphore_giotto,
- show_network = T,
- network_color = 'blue',
- point_size = 1.5,
- cell_color = 'hclust')Lunaphore_giotto <- createSpatialNetwork(Lunaphore_giotto)
+
+spatPlot2D(Lunaphore_giotto,
+ show_network = T,
+ network_color = 'blue',
+ point_size = 1.5,
+ cell_color = 'hclust')
11.4.7 Cell Neighborhood: Cell-Type/Cell-Type Interactions
This is using cellProximityEnrichment() to statistically identify cell type interactions
-cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
- cluster_column = 'cell_types',
- spatial_network_name = 'Delaunay_network',
- adjust_method = 'fdr',
- number_of_simulations = 2000)
-## barplot
-cellProximityBarplot(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- min_orig_ints = 5, min_sim_ints = 5)cell_proximities = cellProximityEnrichment(gobject = Lunaphore_giotto,
+ cluster_column = 'cell_types',
+ spatial_network_name = 'Delaunay_network',
+ adjust_method = 'fdr',
+ number_of_simulations = 2000)
+## barplot
+cellProximityBarplot(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ min_orig_ints = 5, min_sim_ints = 5)
## network
-cellProximityNetwork(gobject = Lunaphore_giotto,
- CPscore = cell_proximities,
- remove_self_edges = T,
- only_show_enrichment_edges = F)## network
+cellProximityNetwork(gobject = Lunaphore_giotto,
+ CPscore = cell_proximities,
+ remove_self_edges = T,
+ only_show_enrichment_edges = F)
9.1 Objective9.2 Background
9.2.1 Visium HD Technology
-
+
9.2.1 Visium HD Technology
Figure 9.1: Overview of Visium HD. Source: 10X Genomics @@ -536,7 +536,7 @@
9.2.1 Visium HD Technology
9.2.2 Colorectal Cancer Sample
-
+
Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics
@@ -550,7 +550,7 @@
9.2.2 Colorectal Cancer Sample9.3 Data Ingestion
9.3.1 Visium HD output data format
-
+
Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline.
@@ -560,19 +560,19 @@
9.3.1 Visium HD output data forma
9.4 Tiling and aggregation
The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells.
-
+
9.5 Scalability and projection functions
diff --git a/visium-part-i.html b/visium-part-i.html
index 5aa1658..c4746ee 100644
--- a/visium-part-i.html
+++ b/visium-part-i.html
@@ -520,7 +520,7 @@ 7 Visium Part I
7.1 The Visium technology
Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression.
-
+
Figure 7.1: Visum workflow. Source: 10X Genomics
@@ -536,73 +536,73 @@
7.2 Introduction to the spatial d
7.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/01_session5/")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
+dir.create("data/01_session5/")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”.
-
+
7.4 Create the Giotto object
createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/01_session5/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-## Provide the path to the visium folder
-data_path <- "data/01_session5/"
-
-## Create object directly from the visium folder
-visium_brain <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/01_session5/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+## Provide the path to the visium folder
+data_path <- "data/01_session5/"
+
+## Create object directly from the visium folder
+visium_brain <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
7.5 Subset on spots that were covered by tissue
Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area.
-spatPlot2D(
- gobject = visium_brain,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE)
-
+spatPlot2D(
+ gobject = visium_brain,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE)
+
Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area.
-
+
7.6 Quality control
@@ -610,43 +610,43 @@ 7.6 Quality controlvisium_brain_statistics <- addStatistics(gobject = visium_brain,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = visium_brain_statistics,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain_statistics <- addStatistics(gobject = visium_brain,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = visium_brain_statistics,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.3: Spatial distribution of features per spot.
filterDistributions() creates a histogram to show the distribution of features per spot across the sample.
-
-
+
+
Figure 7.4: Distribution of features per spot.
When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample.
-
-
+
+
Figure 7.5: Distribution of cells with different features per spot.
filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data:
-filterCombinations(gobject = visium_brain_statistics,
- expression_thresholds = c(1, 2, 3),
- feat_det_in_min_cells = c(50, 100, 200),
- min_det_feats_per_cell = c(500, 1000, 1500))
-
+filterCombinations(gobject = visium_brain_statistics,
+ expression_thresholds = c(1, 2, 3),
+ feat_det_in_min_cells = c(50, 100, 200),
+ min_det_feats_per_cell = c(500, 1000, 1500))
+
Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations.
@@ -656,34 +656,34 @@
7.6 Quality control
7.7 Filtering
Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample.
-visium_brain <- filterGiotto(
- gobject = visium_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE
-)
-Feature type: rna
-Number of cells removed: 4 out of 2702
-Number of feats removed: 7311 out of 22125
+
+
7.8 Normalization
Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one.
-visium_brain <- normalizeGiotto(
- gobject = visium_brain,
- scalefactor = 6000,
- verbose = TRUE
-)
+visium_brain <- normalizeGiotto(
+ gobject = visium_brain,
+ scalefactor = 6000,
+ verbose = TRUE
+)
Calculate the normalized number of features per spot and save the statistics in the metadata table.
-visium_brain <- addStatistics(gobject = visium_brain)
-
-## visualize
-spatPlot2D(gobject = visium_brain,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain <- addStatistics(gobject = visium_brain)
+
+## visualize
+spatPlot2D(gobject = visium_brain,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.7: Spatial distribution of the number of features per spot.
@@ -698,11 +698,11 @@
7.9.1 Highly Variable Features:
loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_loess",
- save_plot = TRUE,
- default_save_name = "HVFplot_loess")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_loess",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_loess")
+
Figure 7.8: Covariance of HVFs using the loess method.
@@ -711,11 +711,11 @@
7.9.1 Highly Variable Features:
pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")
+
Figure 7.9: Variance of HVFs using the pearson residuals method.
@@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
9.2.2 Colorectal Cancer Sample
-
+
9.2.2 Colorectal Cancer Sample9.3 Data Ingestion
Figure 9.2: Colorectal Cancer Overview. Source: 10X Genomics @@ -550,7 +550,7 @@
9.2.2 Colorectal Cancer Sample9.3 Data Ingestion
9.3.1 Visium HD output data format
-
+
Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline.
@@ -560,19 +560,19 @@
9.3.1 Visium HD output data forma
9.4 Tiling and aggregation
The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells.
-
+
9.5 Scalability and projection functions
diff --git a/visium-part-i.html b/visium-part-i.html
index 5aa1658..c4746ee 100644
--- a/visium-part-i.html
+++ b/visium-part-i.html
@@ -520,7 +520,7 @@ 7 Visium Part I
7.1 The Visium technology
Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression.
-
+
Figure 7.1: Visum workflow. Source: 10X Genomics
@@ -536,73 +536,73 @@
7.2 Introduction to the spatial d
7.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/01_session5/")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
+dir.create("data/01_session5/")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”.
-
+
7.4 Create the Giotto object
createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/01_session5/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-## Provide the path to the visium folder
-data_path <- "data/01_session5/"
-
-## Create object directly from the visium folder
-visium_brain <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/01_session5/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+## Provide the path to the visium folder
+data_path <- "data/01_session5/"
+
+## Create object directly from the visium folder
+visium_brain <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
7.5 Subset on spots that were covered by tissue
Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area.
-spatPlot2D(
- gobject = visium_brain,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE)
-
+spatPlot2D(
+ gobject = visium_brain,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE)
+
Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area.
-
+
7.6 Quality control
@@ -610,43 +610,43 @@ 7.6 Quality controlvisium_brain_statistics <- addStatistics(gobject = visium_brain,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = visium_brain_statistics,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain_statistics <- addStatistics(gobject = visium_brain,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = visium_brain_statistics,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.3: Spatial distribution of features per spot.
filterDistributions() creates a histogram to show the distribution of features per spot across the sample.
-
-
+
+
Figure 7.4: Distribution of features per spot.
When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample.
-
-
+
+
Figure 7.5: Distribution of cells with different features per spot.
filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data:
-filterCombinations(gobject = visium_brain_statistics,
- expression_thresholds = c(1, 2, 3),
- feat_det_in_min_cells = c(50, 100, 200),
- min_det_feats_per_cell = c(500, 1000, 1500))
-
+filterCombinations(gobject = visium_brain_statistics,
+ expression_thresholds = c(1, 2, 3),
+ feat_det_in_min_cells = c(50, 100, 200),
+ min_det_feats_per_cell = c(500, 1000, 1500))
+
Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations.
@@ -656,34 +656,34 @@
7.6 Quality control
7.7 Filtering
Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample.
-visium_brain <- filterGiotto(
- gobject = visium_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE
-)
-Feature type: rna
-Number of cells removed: 4 out of 2702
-Number of feats removed: 7311 out of 22125
+
+
7.8 Normalization
Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one.
-visium_brain <- normalizeGiotto(
- gobject = visium_brain,
- scalefactor = 6000,
- verbose = TRUE
-)
+visium_brain <- normalizeGiotto(
+ gobject = visium_brain,
+ scalefactor = 6000,
+ verbose = TRUE
+)
Calculate the normalized number of features per spot and save the statistics in the metadata table.
-visium_brain <- addStatistics(gobject = visium_brain)
-
-## visualize
-spatPlot2D(gobject = visium_brain,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain <- addStatistics(gobject = visium_brain)
+
+## visualize
+spatPlot2D(gobject = visium_brain,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.7: Spatial distribution of the number of features per spot.
@@ -698,11 +698,11 @@
7.9.1 Highly Variable Features:
loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_loess",
- save_plot = TRUE,
- default_save_name = "HVFplot_loess")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_loess",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_loess")
+
Figure 7.8: Covariance of HVFs using the loess method.
@@ -711,11 +711,11 @@
7.9.1 Highly Variable Features:
pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")
+
Figure 7.9: Variance of HVFs using the pearson residuals method.
@@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
9.3.1 Visium HD output data format
-
+
Figure 9.3: File structure of Visium HD data processed with spaceranger pipeline. @@ -560,19 +560,19 @@
9.3.1 Visium HD output data forma
9.4 Tiling and aggregation
The Visium HD data is organized in a grid format. We can aggregate the data into larger bins to reduce the dimensionality of the data. Giotto Suite provides options to bin data not only with squares, but also through hexagons and triangles. Here we use a hexagon tesselation to aggregate the data into arbitrary cells.
- +9.5 Scalability and projection functions
diff --git a/visium-part-i.html b/visium-part-i.html index 5aa1658..c4746ee 100644 --- a/visium-part-i.html +++ b/visium-part-i.html @@ -520,7 +520,7 @@7 Visium Part I
7.1 The Visium technology
Visium allows you to perform spatial transcriptomics, which combines histological information with whole transcriptome gene expression profiles (fresh frozen or FFPE) to provide you with spatially resolved gene expression.
-
+
Figure 7.1: Visum workflow. Source: 10X Genomics
@@ -536,73 +536,73 @@
7.2 Introduction to the spatial d
7.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/01_session5/")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
+dir.create("data/01_session5/")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”.
-
+
7.4 Create the Giotto object
createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/01_session5/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-## Provide the path to the visium folder
-data_path <- "data/01_session5/"
-
-## Create object directly from the visium folder
-visium_brain <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/01_session5/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+## Provide the path to the visium folder
+data_path <- "data/01_session5/"
+
+## Create object directly from the visium folder
+visium_brain <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
7.5 Subset on spots that were covered by tissue
Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area.
-spatPlot2D(
- gobject = visium_brain,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE)
-
+spatPlot2D(
+ gobject = visium_brain,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE)
+
Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area.
-
+
7.6 Quality control
@@ -610,43 +610,43 @@ 7.6 Quality controlvisium_brain_statistics <- addStatistics(gobject = visium_brain,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = visium_brain_statistics,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain_statistics <- addStatistics(gobject = visium_brain,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = visium_brain_statistics,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.3: Spatial distribution of features per spot.
filterDistributions() creates a histogram to show the distribution of features per spot across the sample.
-
-
+
+
Figure 7.4: Distribution of features per spot.
When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample.
-
-
+
+
Figure 7.5: Distribution of cells with different features per spot.
filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data:
-filterCombinations(gobject = visium_brain_statistics,
- expression_thresholds = c(1, 2, 3),
- feat_det_in_min_cells = c(50, 100, 200),
- min_det_feats_per_cell = c(500, 1000, 1500))
-
+filterCombinations(gobject = visium_brain_statistics,
+ expression_thresholds = c(1, 2, 3),
+ feat_det_in_min_cells = c(50, 100, 200),
+ min_det_feats_per_cell = c(500, 1000, 1500))
+
Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations.
@@ -656,34 +656,34 @@
7.6 Quality control
7.7 Filtering
Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample.
-visium_brain <- filterGiotto(
- gobject = visium_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE
-)
-Feature type: rna
-Number of cells removed: 4 out of 2702
-Number of feats removed: 7311 out of 22125
+
+
7.8 Normalization
Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one.
-visium_brain <- normalizeGiotto(
- gobject = visium_brain,
- scalefactor = 6000,
- verbose = TRUE
-)
+visium_brain <- normalizeGiotto(
+ gobject = visium_brain,
+ scalefactor = 6000,
+ verbose = TRUE
+)
Calculate the normalized number of features per spot and save the statistics in the metadata table.
-visium_brain <- addStatistics(gobject = visium_brain)
-
-## visualize
-spatPlot2D(gobject = visium_brain,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain <- addStatistics(gobject = visium_brain)
+
+## visualize
+spatPlot2D(gobject = visium_brain,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.7: Spatial distribution of the number of features per spot.
@@ -698,11 +698,11 @@
7.9.1 Highly Variable Features:
loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_loess",
- save_plot = TRUE,
- default_save_name = "HVFplot_loess")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_loess",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_loess")
+
Figure 7.8: Covariance of HVFs using the loess method.
@@ -711,11 +711,11 @@
7.9.1 Highly Variable Features:
pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")
+
Figure 7.9: Variance of HVFs using the pearson residuals method.
@@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
Figure 7.1: Visum workflow. Source: 10X Genomics @@ -536,73 +536,73 @@
7.2 Introduction to the spatial d
7.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/01_session5/")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
+dir.create("data/01_session5/")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")
After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”.
-
+
7.4 Create the Giotto object
createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/01_session5/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-## Provide the path to the visium folder
-data_path <- "data/01_session5/"
-
-## Create object directly from the visium folder
-visium_brain <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)
+library(Giotto)
+
+## Set instructions
+results_folder <- "results/01_session5/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+## Provide the path to the visium folder
+data_path <- "data/01_session5/"
+
+## Create object directly from the visium folder
+visium_brain <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)
7.5 Subset on spots that were covered by tissue
Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area.
-spatPlot2D(
- gobject = visium_brain,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE)
-
+spatPlot2D(
+ gobject = visium_brain,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE)
+
Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area.
-
+
7.6 Quality control
@@ -610,43 +610,43 @@ 7.6 Quality controlvisium_brain_statistics <- addStatistics(gobject = visium_brain,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = visium_brain_statistics,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain_statistics <- addStatistics(gobject = visium_brain,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = visium_brain_statistics,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.3: Spatial distribution of features per spot.
filterDistributions() creates a histogram to show the distribution of features per spot across the sample.
-
-
+
+
Figure 7.4: Distribution of features per spot.
When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample.
-
-
+
+
Figure 7.5: Distribution of cells with different features per spot.
filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data:
-filterCombinations(gobject = visium_brain_statistics,
- expression_thresholds = c(1, 2, 3),
- feat_det_in_min_cells = c(50, 100, 200),
- min_det_feats_per_cell = c(500, 1000, 1500))
-
+filterCombinations(gobject = visium_brain_statistics,
+ expression_thresholds = c(1, 2, 3),
+ feat_det_in_min_cells = c(50, 100, 200),
+ min_det_feats_per_cell = c(500, 1000, 1500))
+
Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations.
@@ -656,34 +656,34 @@
7.6 Quality control
7.7 Filtering
Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample.
-visium_brain <- filterGiotto(
- gobject = visium_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE
-)
-Feature type: rna
-Number of cells removed: 4 out of 2702
-Number of feats removed: 7311 out of 22125
+
+
7.8 Normalization
Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one.
-visium_brain <- normalizeGiotto(
- gobject = visium_brain,
- scalefactor = 6000,
- verbose = TRUE
-)
+visium_brain <- normalizeGiotto(
+ gobject = visium_brain,
+ scalefactor = 6000,
+ verbose = TRUE
+)
Calculate the normalized number of features per spot and save the statistics in the metadata table.
-visium_brain <- addStatistics(gobject = visium_brain)
-
-## visualize
-spatPlot2D(gobject = visium_brain,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
-
+visium_brain <- addStatistics(gobject = visium_brain)
+
+## visualize
+spatPlot2D(gobject = visium_brain,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)
+
Figure 7.7: Spatial distribution of the number of features per spot.
@@ -698,11 +698,11 @@
7.9.1 Highly Variable Features:
loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_loess",
- save_plot = TRUE,
- default_save_name = "HVFplot_loess")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_loess",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_loess")
+
Figure 7.8: Covariance of HVFs using the loess method.
@@ -711,11 +711,11 @@
7.9.1 Highly Variable Features:
pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")
+
Figure 7.9: Variance of HVFs using the pearson residuals method.
@@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
7.3 Download dataset
You need to download the expression matrix and spatial information by running these commands:
-dir.create("data/01_session5/")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
-
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
- destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")dir.create("data/01_session5/")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_raw_feature_bc_matrix.tar.gz")
+
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.1.0/V1_Adult_Mouse_Brain/V1_Adult_Mouse_Brain_spatial.tar.gz",
+ destfile = "data/01_session5/V1_Adult_Mouse_Brain_spatial.tar.gz")After downloading, unzip the gz files. You should get the “raw_feature_bc_matrix” and “spatial” folders inside “data/01_session5/”.
- +7.4 Create the Giotto object
createGiottoVisiumObject() will look for the standardized files organization from the visium technology in the data folder and will automatically load the expression and spatial information to create the Giotto object.
-library(Giotto)
-
-## Set instructions
-results_folder <- "results/01_session5/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- return_plot = FALSE,
- python_path = python_path
-)
-
-## Provide the path to the visium folder
-data_path <- "data/01_session5/"
-
-## Create object directly from the visium folder
-visium_brain <- createGiottoVisiumObject(
- visium_dir = data_path,
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instructions
-)library(Giotto)
+
+## Set instructions
+results_folder <- "results/01_session5/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ return_plot = FALSE,
+ python_path = python_path
+)
+
+## Provide the path to the visium folder
+data_path <- "data/01_session5/"
+
+## Create object directly from the visium folder
+visium_brain <- createGiottoVisiumObject(
+ visium_dir = data_path,
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instructions
+)7.5 Subset on spots that were covered by tissue
Use the metadata column “in_tissue” to highlight the spots corresponding to the tissue area.
-spatPlot2D(
- gobject = visium_brain,
- cell_color = "in_tissue",
- point_size = 2,
- cell_color_code = c("0" = "lightgrey", "1" = "blue"),
- show_image = TRUE)
+
+
spatPlot2D(
+ gobject = visium_brain,
+ cell_color = "in_tissue",
+ point_size = 2,
+ cell_color_code = c("0" = "lightgrey", "1" = "blue"),
+ show_image = TRUE)Figure 7.2: Spatial plot of the Visium mouse brain sample, color indicates wheter the spot is in tissue (1) or not (0).
Use the same metadata column “in_tissue” to subset the object and keep only the spots corresponding to the tissue area.
- +7.6 Quality control
@@ -610,43 +610,43 @@7.6 Quality controlvisium_brain_statistics <- addStatistics(gobject = visium_brain,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = visium_brain_statistics,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
visium_brain_statistics <- addStatistics(gobject = visium_brain,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = visium_brain_statistics,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
+
+
visium_brain_statistics <- addStatistics(gobject = visium_brain,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = visium_brain_statistics,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)Figure 7.3: Spatial distribution of features per spot.
filterDistributions() creates a histogram to show the distribution of features per spot across the sample.
- -
+
+
Figure 7.4: Distribution of features per spot.
When setting the detection = “feats”, the histogram shows the distribution of cells with certain numbers of features across the sample.
- -
+
+
-
Figure 7.5: Distribution of cells with different features per spot.
filterCombinations() may be used to test how different filtering parameters will affect the number of cells and features in the filtered data:
-filterCombinations(gobject = visium_brain_statistics,
- expression_thresholds = c(1, 2, 3),
- feat_det_in_min_cells = c(50, 100, 200),
- min_det_feats_per_cell = c(500, 1000, 1500))
+
+
7.6 Quality control
+
-
filterCombinations(gobject = visium_brain_statistics,
+ expression_thresholds = c(1, 2, 3),
+ feat_det_in_min_cells = c(50, 100, 200),
+ min_det_feats_per_cell = c(500, 1000, 1500))Figure 7.6: Number of spots and features filtered when using multiple feat_det_in_min_cells and min_det_feats_per_cell combinations. @@ -656,34 +656,34 @@
7.6 Quality control
7.7 Filtering
Use the arguments feat_det_in_min_cells and min_det_feats_per_cell to set the minimal number of cells where an individual feature must be detected and the minimal number of features per spot/cell, respectively, to filter the giotto object. All the features and cells under those thresholds will be removed from the sample.
-visium_brain <- filterGiotto(
- gobject = visium_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE
-)
-Feature type: rna
-Number of cells removed: 4 out of 2702
-Number of feats removed: 7311 out of 22125
+
+
visium_brain <- filterGiotto(
- gobject = visium_brain,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 1000,
- expression_values = "raw",
- verbose = TRUE
-)Feature type: rna
-Number of cells removed: 4 out of 2702
-Number of feats removed: 7311 out of 22125 7.8 Normalization
Use scalefactor to set the scale factor to use after library size normalization. The default value is 6000, but you can use a different one.
-visium_brain <- normalizeGiotto(
- gobject = visium_brain,
- scalefactor = 6000,
- verbose = TRUE
-)visium_brain <- normalizeGiotto(
+ gobject = visium_brain,
+ scalefactor = 6000,
+ verbose = TRUE
+)Calculate the normalized number of features per spot and save the statistics in the metadata table.
-visium_brain <- addStatistics(gobject = visium_brain)
-
-## visualize
-spatPlot2D(gobject = visium_brain,
- cell_color = "nr_feats",
- color_as_factor = FALSE)
+
+
7.9.1 Highly Variable Features:
visium_brain <- addStatistics(gobject = visium_brain)
+
+## visualize
+spatPlot2D(gobject = visium_brain,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE)Figure 7.7: Spatial distribution of the number of features per spot. @@ -698,11 +698,11 @@
7.9.1 Highly Variable Features:
loess regression is used when the relationship between mean expression and variance is non-linear or can be described by a non-parametric model.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_loess",
- save_plot = TRUE,
- default_save_name = "HVFplot_loess")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_loess",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_loess")
+
Figure 7.8: Covariance of HVFs using the loess method.
@@ -711,11 +711,11 @@
7.9.1 Highly Variable Features:
pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")
+
Figure 7.9: Variance of HVFs using the pearson residuals method.
@@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_loess",
- save_plot = TRUE,
- default_save_name = "HVFplot_loess")visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_loess",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_loess")Figure 7.8: Covariance of HVFs using the loess method. @@ -711,11 +711,11 @@
7.9.1 Highly Variable Features:
pearson residuals are used for variance stabilization (to account for technical noise) and highlighting overdispersed genes.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")
+
Figure 7.9: Variance of HVFs using the pearson residuals method.
@@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
visium_brain <- calculateHVF(gobject = visium_brain,
- method = "var_p_resid",
- save_plot = TRUE,
- default_save_name = "HVFplot_pearson")visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "var_p_resid",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_pearson")Figure 7.9: Variance of HVFs using the pearson residuals method. @@ -724,11 +724,11 @@
7.9.1 Highly Variable Features:
binned (covariance groups) are used when gene expression variability differs across expression levels or spatial regions, without assuming a specific relationship between mean expression and variance. This is the default method in the calculateHVF() function.
-visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")
-
+visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")
+
Figure 7.10: Covariance of HVFs using the binned method.
@@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
+
- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
-my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")
+my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")
- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
-
-
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
-
-
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
-
-
+
+
Figure 7.13: PCA using custom features.
@@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
visium_brain <- calculateHVF(gobject = visium_brain,
- method = "cov_groups",
- save_plot = TRUE,
- default_save_name = "HVFplot_binned")visium_brain <- calculateHVF(gobject = visium_brain,
+ method = "cov_groups",
+ save_plot = TRUE,
+ default_save_name = "HVFplot_binned")Figure 7.10: Covariance of HVFs using the binned method. @@ -744,41 +744,41 @@
7.10.1 PCAvisium_brain <- runPCA(gobject = visium_brain)
visium_brain <- runPCA(gobject = visium_brain)- You can also use specific features for the Principal Components calculation, by passing a vector of features in the “feats_to_use” argument.
my_features <- head(getFeatureMetadata(visium_brain,
- output = "data.table")$feat_ID,
- 1000)
-
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = my_features,
- name = "custom_pca")my_features <- head(getFeatureMetadata(visium_brain,
+ output = "data.table")$feat_ID,
+ 1000)
+
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = my_features,
+ name = "custom_pca")- Visualization
Create a screeplot to visualize the percentage of variance explained by each component.
- -
+
+
Figure 7.11: Screeplot showing the variance explained per principal component.
Visualized the PCA calculated using the HVFs.
- -
+
+
Figure 7.12: PCA plot using HVFs.
Visualized the custom PCA calculated using the vector of features.
- -
+
+
7.10.1 PCA
Figure 7.13: PCA using custom features. @@ -788,13 +788,13 @@
7.10.1 PCA
7.10.2 UMAP
-
+
- Visualization
-
-
+
+
Figure 7.14: UMAP using the 10 first principal components.
@@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
Figure 7.14: UMAP using the 10 first principal components. @@ -803,13 +803,13 @@
7.10.2 UMAP
7.10.3 t-SNE
-
+
- Visualization
-
-
+
+
Figure 7.15: tSNE using the 10 first principal components.
@@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
+
- Create a kNN network
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")
7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
-
+
- Visualization
-
-
+
+
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)
+
Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
-
+plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)
+
Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
-
-
+
+
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
-
+plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)
+
Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
-
-
+
+
Figure 7.21: Spatial plot, colors indicate the Leiden clusters.
@@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))
+
Figure 8.3: Spatial distribution of the top gini genes scaled expression.
@@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
+scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
-
+violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))
+
Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
-
+plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)
+
Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
-
+dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))
+
Figure 8.6: Spatial distribution of the top scran genes scaled expression.
@@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
Figure 7.15: tSNE using the 10 first principal components. @@ -822,81 +822,81 @@
7.11 Clusteringvisium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)
visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15)- Create a kNN network
visium_brain <- createNearestNetwork(gobject = visium_brain,
- dimensions_to_use = 1:10,
- k = 15,
- type = "kNN")visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dimensions_to_use = 1:10,
+ k = 15,
+ type = "kNN")7.11.1 Calculate Leiden clustering
Use the previously calculated shared nearest neighbors to create clusters. The default resolution is 1, but you can decrease the value to avoid the over calculation of clusters.
- +- Visualization
+
+
-
Figure 7.16: PCA plot, colors indicate the Leiden clusters.
Use the cluster IDs to visualize the clusters in the UMAP space.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = FALSE,
- point_size = 2.5)
+
+
-
plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = FALSE,
+ point_size = 2.5)Figure 7.17: UMAP plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotUMAP(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_NN_network = TRUE,
- point_size = 2.5)
+
+
plotUMAP(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_NN_network = TRUE,
+ point_size = 2.5)Figure 7.18: UMAP showing the nearest network.
Use the cluster IDs to visualize the clusters on the tSNE.
- -
+
+
-
Figure 7.19: tSNE plot, colors indicate the Leiden clusters.
Set the argument “show_NN_network = TRUE” to visualize the connections between spots.
-plotTSNE(gobject = visium_brain,
- cell_color = "leiden_clus",
- point_size = 2.5,
- show_NN_network = TRUE)
+
+
plotTSNE(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ point_size = 2.5,
+ show_NN_network = TRUE)Figure 7.20: tSNE showing the nearest network.
Use the cluster IDs to visualize their spatial location.
- -
+
+
7.11.1 Calculate Leiden clusterin
-
Figure 7.21: Spatial plot, colors indicate the Leiden clusters. @@ -906,10 +906,10 @@
7.11.1 Calculate Leiden clusterin
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
-
-
-
+
+
+
Figure 7.22: Spatial plot, colors indicate the Louvain clusters.
@@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@ 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@ 8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
+gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
- Visualize
7.11.2 Calculate Louvain clustering
Louvain is an alternative clustering method, used to detect communities in large networks.
- - -
+
+
+
+
+
diff --git a/visium-part-ii.html b/visium-part-ii.html
index 5cad4aa..734c8ee 100644
--- a/visium-part-ii.html
+++ b/visium-part-ii.html
@@ -519,9 +519,9 @@
Figure 7.22: Spatial plot, colors indicate the Louvain clusters. @@ -920,89 +920,89 @@
7.11.2 Calculate Louvain clusteri
7.13 Session info
- -R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] colorRamp2_0.1.0 deldir_2.0-4
- [3] rlang_1.1.4 magrittr_2.0.3
- [5] GiottoUtils_0.1.10 matrixStats_1.3.0
- [7] compiler_4.4.1 png_0.1-8
- [9] systemfonts_1.1.0 vctrs_0.6.5
- [11] reshape2_1.4.4 stringr_1.5.1
- [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
- [15] crayon_1.5.3 fastmap_1.2.0
- [17] backports_1.5.0 magick_2.8.4
- [19] XVector_0.44.0 labeling_0.4.3
- [21] utf8_1.2.4 rmarkdown_2.27
- [23] UCSC.utils_1.0.0 ragg_1.3.2
- [25] purrr_1.0.2 xfun_0.46
- [27] beachmat_2.20.0 zlibbioc_1.50.0
- [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
- [31] DelayedArray_0.30.1 BiocParallel_1.38.0
- [33] terra_1.7-78 irlba_2.3.5.1
- [35] parallel_4.4.1 R6_2.5.1
- [37] stringi_1.8.4 RColorBrewer_1.1-3
- [39] reticulate_1.38.0 parallelly_1.37.1
- [41] GenomicRanges_1.56.1 scattermore_1.2
- [43] Rcpp_1.0.13 bookdown_0.40
- [45] SummarizedExperiment_1.34.0 knitr_1.48
- [47] future.apply_1.11.2 R.utils_2.12.3
- [49] FNN_1.1.4 IRanges_2.38.1
- [51] Matrix_1.7-0 igraph_2.0.3
- [53] tidyselect_1.2.1 rstudioapi_0.16.0
- [55] abind_1.4-5 yaml_2.3.9
- [57] codetools_0.2-20 listenv_0.9.1
- [59] lattice_0.22-6 tibble_3.2.1
- [61] plyr_1.8.9 Biobase_2.64.0
- [63] withr_3.0.0 Rtsne_0.17
- [65] evaluate_0.24.0 future_1.33.2
- [67] pillar_1.9.0 MatrixGenerics_1.16.0
- [69] checkmate_2.3.1 stats4_4.4.1
- [71] plotly_4.10.4 generics_0.1.3
- [73] dbscan_1.2-0 sp_2.1-4
- [75] S4Vectors_0.42.1 ggplot2_3.5.1
- [77] munsell_0.5.1 scales_1.3.0
- [79] globals_0.16.3 gtools_3.9.5
- [81] glue_1.7.0 lazyeval_0.2.2
- [83] tools_4.4.1 GiottoVisuals_0.2.4
- [85] data.table_1.15.4 ScaledMatrix_1.12.0
- [87] cowplot_1.1.3 grid_4.4.1
- [89] tidyr_1.3.1 colorspace_2.1-0
- [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
- [93] BiocSingular_1.20.0 rsvd_1.0.5
- [95] cli_3.6.3 textshaping_0.4.0
- [97] fansi_1.0.6 S4Arrays_1.4.1
- [99] viridisLite_0.4.2 dplyr_1.1.4
-[101] uwot_0.2.2 gtable_0.3.5
-[103] R.methodsS3_1.8.2 digest_0.6.36
-[105] BiocGenerics_0.50.0 SparseArray_1.4.8
-[107] ggrepel_0.9.5 farver_2.1.2
-[109] rjson_0.2.21 htmlwidgets_1.6.4
-[111] htmltools_0.5.8.1 R.oo_1.26.0
-[113] lifecycle_1.0.4 httr_1.4.7 R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] colorRamp2_0.1.0 deldir_2.0-4
+ [3] rlang_1.1.4 magrittr_2.0.3
+ [5] GiottoUtils_0.1.10 matrixStats_1.3.0
+ [7] compiler_4.4.1 png_0.1-8
+ [9] systemfonts_1.1.0 vctrs_0.6.5
+ [11] reshape2_1.4.4 stringr_1.5.1
+ [13] pkgconfig_2.0.3 SpatialExperiment_1.14.0
+ [15] crayon_1.5.3 fastmap_1.2.0
+ [17] backports_1.5.0 magick_2.8.4
+ [19] XVector_0.44.0 labeling_0.4.3
+ [21] utf8_1.2.4 rmarkdown_2.27
+ [23] UCSC.utils_1.0.0 ragg_1.3.2
+ [25] purrr_1.0.2 xfun_0.46
+ [27] beachmat_2.20.0 zlibbioc_1.50.0
+ [29] GenomeInfoDb_1.40.1 jsonlite_1.8.8
+ [31] DelayedArray_0.30.1 BiocParallel_1.38.0
+ [33] terra_1.7-78 irlba_2.3.5.1
+ [35] parallel_4.4.1 R6_2.5.1
+ [37] stringi_1.8.4 RColorBrewer_1.1-3
+ [39] reticulate_1.38.0 parallelly_1.37.1
+ [41] GenomicRanges_1.56.1 scattermore_1.2
+ [43] Rcpp_1.0.13 bookdown_0.40
+ [45] SummarizedExperiment_1.34.0 knitr_1.48
+ [47] future.apply_1.11.2 R.utils_2.12.3
+ [49] FNN_1.1.4 IRanges_2.38.1
+ [51] Matrix_1.7-0 igraph_2.0.3
+ [53] tidyselect_1.2.1 rstudioapi_0.16.0
+ [55] abind_1.4-5 yaml_2.3.9
+ [57] codetools_0.2-20 listenv_0.9.1
+ [59] lattice_0.22-6 tibble_3.2.1
+ [61] plyr_1.8.9 Biobase_2.64.0
+ [63] withr_3.0.0 Rtsne_0.17
+ [65] evaluate_0.24.0 future_1.33.2
+ [67] pillar_1.9.0 MatrixGenerics_1.16.0
+ [69] checkmate_2.3.1 stats4_4.4.1
+ [71] plotly_4.10.4 generics_0.1.3
+ [73] dbscan_1.2-0 sp_2.1-4
+ [75] S4Vectors_0.42.1 ggplot2_3.5.1
+ [77] munsell_0.5.1 scales_1.3.0
+ [79] globals_0.16.3 gtools_3.9.5
+ [81] glue_1.7.0 lazyeval_0.2.2
+ [83] tools_4.4.1 GiottoVisuals_0.2.4
+ [85] data.table_1.15.4 ScaledMatrix_1.12.0
+ [87] cowplot_1.1.3 grid_4.4.1
+ [89] tidyr_1.3.1 colorspace_2.1-0
+ [91] SingleCellExperiment_1.26.0 GenomeInfoDbData_1.2.12
+ [93] BiocSingular_1.20.0 rsvd_1.0.5
+ [95] cli_3.6.3 textshaping_0.4.0
+ [97] fansi_1.0.6 S4Arrays_1.4.1
+ [99] viridisLite_0.4.2 dplyr_1.1.4
+[101] uwot_0.2.2 gtable_0.3.5
+[103] R.methodsS3_1.8.2 digest_0.6.36
+[105] BiocGenerics_0.50.0 SparseArray_1.4.8
+[107] ggrepel_0.9.5 farver_2.1.2
+[109] rjson_0.2.21 htmlwidgets_1.6.4
+[111] htmltools_0.5.8.1 R.oo_1.26.0
+[113] lifecycle_1.0.4 httr_1.4.7 8 Visium Part II
8.1 Load the object
-
+
8.2 Differential expression
@@ -531,48 +531,48 @@8.2.1 Gini markersgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$feats
gini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "gini",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$featsgini_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "gini",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_gini <- gini_markers[, head(.SD, 2), by = "cluster"]$featsPlot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_gini),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
+
+
-
violinPlot(visium_brain,
+ feats = unique(topgenes_gini),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))Figure 8.1: Violin plot showing the top gini genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_gini),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
+
+
-
plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_gini),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)Figure 8.2: Heatmap showing the top gini genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_gini)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 15, base_height = 20))
+
+
8.2.2 Scran markers
+
-
dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_gini)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 15, base_height = 20))Figure 8.3: Spatial distribution of the top gini genes scaled expression. @@ -585,48 +585,48 @@
8.2.2 Scran markersscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats
scran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "leiden_clus",
- min_feats = 10)
-
-topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$featsscran_markers <- findMarkers_one_vs_all(gobject = visium_brain,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "leiden_clus",
+ min_feats = 10)
+
+topgenes_scran <- scran_markers[, head(.SD, 2), by = "cluster"]$feats- Visualize
Plot the normalized expression distribution of the top expressed genes.
-violinPlot(visium_brain,
- feats = unique(topgenes_scran),
- cluster_column = "leiden_clus",
- strip_text = 6,
- strip_position = "right",
- save_param = list(base_width = 5, base_height = 30))
+
+
-
violinPlot(visium_brain,
+ feats = unique(topgenes_scran),
+ cluster_column = "leiden_clus",
+ strip_text = 6,
+ strip_position = "right",
+ save_param = list(base_width = 5, base_height = 30))Figure 8.4: Violin plot of the top scran genes normalized expression.
Use the cluster IDs to create a heatmap with the normalized expression of the top expressed genes per cluster.
-plotMetaDataHeatmap(visium_brain,
- selected_feats = unique(topgenes_scran),
- metadata_cols = "leiden_clus",
- x_text_size = 10, y_text_size = 10)
+
+
-
plotMetaDataHeatmap(visium_brain,
+ selected_feats = unique(topgenes_scran),
+ metadata_cols = "leiden_clus",
+ x_text_size = 10, y_text_size = 10)Figure 8.5: Heatmap showing the top scran genes normalized expression per Leiden cluster.
Visualize the scaled expression spatial distribution of the top expressed genes across the sample.
-dimFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = sort(unique(topgenes_scran)),
- cow_n_col = 5,
- point_size = 1,
- save_param = list(base_width = 20, base_height = 20))
+
+
8.3 Enrichment & Deconvolutio
dimFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = sort(unique(topgenes_scran)),
+ cow_n_col = 5,
+ point_size = 1,
+ save_param = list(base_width = 20, base_height = 20))Figure 8.6: Spatial distribution of the top scran genes scaled expression. @@ -641,89 +641,89 @@
8.3 Enrichment & Deconvolutio
- Download the single-cell dataset
-
+
- Create the single-cell object and run the normalization step
-results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)
+results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)
8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
-markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster))
+markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster))
- Create the signature matrix
-sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)
+sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)
- Run the enrichment test with PAGE
-
+
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
-
+cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)
+
Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
-
+spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)
+
Figure 8.8: Spatial distribution of cell types identified using the PAGE method.
@@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
+sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)
- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
-
+
- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
-
+spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method.
@@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
results_folder <- "results/02_session1/"
-
-python_path <- NULL
-
-instructions <- createGiottoInstructions(
- save_dir = results_folder,
- save_plot = TRUE,
- show_plot = FALSE,
- python_path = python_path
-)
-
-sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
-sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
-
-giotto_SC <- createGiottoObject(expression = sc_expression,
- instructions = instructions)
-
-giotto_SC <- addCellMetadata(giotto_SC,
- new_metadata = data.table::fread(sc_metadata))
-
-giotto_SC <- normalizeGiotto(giotto_SC)results_folder <- "results/02_session1/"
+
+python_path <- NULL
+
+instructions <- createGiottoInstructions(
+ save_dir = results_folder,
+ save_plot = TRUE,
+ show_plot = FALSE,
+ python_path = python_path
+)
+
+sc_expression <- "data/02_session1/brain_sc_expression_matrix.txt.gz"
+sc_metadata <- "data/02_session1/brain_sc_metadata.csv"
+
+giotto_SC <- createGiottoObject(expression = sc_expression,
+ instructions = instructions)
+
+giotto_SC <- addCellMetadata(giotto_SC,
+ new_metadata = data.table::fread(sc_metadata))
+
+giotto_SC <- normalizeGiotto(giotto_SC)8.3.1 PAGE/Rank
Parametric Analysis of Gene Set Enrichment (PAGE) and Rank enrichment both aim to determine whether a predefined set of genes show statistically significant differences in expression compared to other genes in the dataset.
- Calculate the cell type markers
markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
- method = "scran",
- expression_values = "normalized",
- cluster_column = "Class",
- min_feats = 3)
-
-top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
-celltypes <- levels(factor(markers_scran$cluster)) markers_scran <- findMarkers_one_vs_all(gobject = giotto_SC,
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "Class",
+ min_feats = 3)
+
+top_markers <- markers_scran[, head(.SD, 10), by = "cluster"]
+celltypes <- levels(factor(markers_scran$cluster)) - Create the signature matrix
sign_list <- list()
-
-for (i in 1:length(celltypes)){
- sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
-}
-
-sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
- sign_list = sign_list)sign_list <- list()
+
+for (i in 1:length(celltypes)){
+ sign_list[[i]] = top_markers[which(top_markers$cluster == celltypes[i]),]$feats
+}
+
+sign_matrix <- makeSignMatrixPAGE(sign_names = celltypes,
+ sign_list = sign_list)- Run the enrichment test with PAGE
- Visualize
Create a heatmap showing the enrichment of cell types (from the single-cell data annotation) in the spatial dataset clusters.
-cell_types_PAGE <- colnames(sign_matrix)
-
-plotMetaDataCellsHeatmap(gobject = visium_brain,
- metadata_cols = "leiden_clus",
- value_cols = cell_types_PAGE,
- spat_enr_names = "PAGE",
- x_text_size = 8,
- y_text_size = 8)
+
+
-
cell_types_PAGE <- colnames(sign_matrix)
+
+plotMetaDataCellsHeatmap(gobject = visium_brain,
+ metadata_cols = "leiden_clus",
+ value_cols = cell_types_PAGE,
+ spat_enr_names = "PAGE",
+ x_text_size = 8,
+ y_text_size = 8)Figure 8.7: Cell types enrichment per Leiden cluster, identified using the PAGE method.
Plot the spatial distribution of the cell types.
-spatCellPlot2D(gobject = visium_brain,
- spat_enr_names = "PAGE",
- cell_annotation_values = cell_types_PAGE,
- cow_n_col = 3,
- coord_fix_ratio = 1,
- point_size = 1,
- show_legend = TRUE)
+
+
8.3.2 SpatialDWLS
+
-
spatCellPlot2D(gobject = visium_brain,
+ spat_enr_names = "PAGE",
+ cell_annotation_values = cell_types_PAGE,
+ cow_n_col = 3,
+ coord_fix_ratio = 1,
+ point_size = 1,
+ show_legend = TRUE)Figure 8.8: Spatial distribution of cell types identified using the PAGE method. @@ -736,27 +736,27 @@
8.3.2 SpatialDWLSsign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)
sign_matrix <- makeSignMatrixDWLSfromMatrix(
- matrix = getExpression(giotto_SC,
- values = "normalized",
- output = "matrix"),
- cell_type = pDataDT(giotto_SC)$Class,
- sign_gene = top_markers$feats)sign_matrix <- makeSignMatrixDWLSfromMatrix(
+ matrix = getExpression(giotto_SC,
+ values = "normalized",
+ output = "matrix"),
+ cell_type = pDataDT(giotto_SC)$Class,
+ sign_gene = top_markers$feats)- Run the DWLS Deconvolution
This step may take a couple of minutes to run.
- +- Visualize
Plot the DWLS deconvolution result creating with pie plots showing the proportion of each cell type per spot.
-spatDeconvPlot(visium_brain,
- show_image = FALSE,
- radius = 50,
- save_param = list(save_name = "8_spat_DWLS_pie_plot"))
+
+
8.4.1 Spatial variable genes
spatDeconvPlot(visium_brain,
+ show_image = FALSE,
+ radius = 50,
+ save_param = list(save_name = "8_spat_DWLS_pie_plot"))Figure 8.9: Spatial deconvolution plot showing the proportion of cell types per spot, identified using the DWLS method. @@ -771,17 +771,17 @@
8.4.1 Spatial variable genes
Create a spatial network
-visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")
-
+visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")
+
Figure 8.10: Spatial network across spots in the Visium mouse sample.
@@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
visium_brain <- createSpatialNetwork(gobject = visium_brain,
- method = "kNN",
- k = 6,
- maximum_distance_knn = 400,
- name = "spatial_network")
-
-spatPlot2D(gobject = visium_brain,
- show_network= TRUE,
- network_color = "blue",
- spatial_network_name = "spatial_network")visium_brain <- createSpatialNetwork(gobject = visium_brain,
+ method = "kNN",
+ k = 6,
+ maximum_distance_knn = 400,
+ name = "spatial_network")
+
+spatPlot2D(gobject = visium_brain,
+ show_network= TRUE,
+ network_color = "blue",
+ spatial_network_name = "spatial_network")Figure 8.10: Spatial network across spots in the Visium mouse sample. @@ -792,21 +792,21 @@
8.4.1 Spatial variable genes
Rank the genes on the spatial dataset depending on whether they exhibit a spatial pattern location or not.
This step may take a few minutes to run.
-ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")
+ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")
- Visualize top results
Plot the scaled expression of genes with the highest probability of being spatial genes.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)
+
Figure 8.11: Spatial distribution of the top spatial genes scaled expression.
@@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)
+
Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp.
@@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
-
+heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))
+
Figure 8.13: Correlations heatmap between spatial genes and correlated clusters.
@@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
-
+top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)
+
Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group.
@@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
-
+spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)
+
Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster.
@@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
ranktest <- binSpect(visium_brain,
- bin_method = "rank",
- calc_hub = TRUE,
- hub_min_int = 5,
- spatial_network_name = "spatial_network")ranktest <- binSpect(visium_brain,
+ bin_method = "rank",
+ calc_hub = TRUE,
+ hub_min_int = 5,
+ spatial_network_name = "spatial_network")spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = ranktest$feats[1:6],
- cow_n_col = 2,
- point_size = 1)spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = ranktest$feats[1:6],
+ cow_n_col = 2,
+ point_size = 1)Figure 8.11: Spatial distribution of the top spatial genes scaled expression. @@ -818,30 +818,30 @@
8.4.2 Spatial co-expression modul
- Cluster the top 500 spatial genes into 20 clusters
-
+
- Use detectSpatialCorGenes function to calculate pairwise distances between genes.
-spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)
+spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)
- Identify most similar spatially correlated genes for one gene
-
+
- Visualize
spat_cor_netw_DT <- detectSpatialCorFeats(
- visium_brain,
- method = "network",
- spatial_network_name = "spatial_network",
- subset_feats = ext_spatial_genes)spat_cor_netw_DT <- detectSpatialCorFeats(
+ visium_brain,
+ method = "network",
+ spatial_network_name = "spatial_network",
+ subset_feats = ext_spatial_genes)Plot the scaled expression of the 3 genes with most similar spatial patterns to Mbp.
-spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- feats = top10_genes$variable[1:4],
- point_size = 1.5)
+
+
8.4.2 Spatial co-expression modul
-
spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ feats = top10_genes$variable[1:4],
+ point_size = 1.5)Figure 8.12: Spatial distribution of the scaled expression of 3 genes with similar spatial pattern to Mbp. @@ -850,18 +850,18 @@
8.4.2 Spatial co-expression modul
- Cluster spatial genes
-
+
- Visualize clusters
Plot the correlation of the top 500 spatial genes with their assigned cluster.
-heatmSpatialCorFeats(visium_brain,
- spatCorObject = spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- heatmap_legend_param = list(title = NULL))
+
+
8.4.2 Spatial co-expression modul
-
heatmSpatialCorFeats(visium_brain,
+ spatCorObject = spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ heatmap_legend_param = list(title = NULL))Figure 8.13: Correlations heatmap between spatial genes and correlated clusters. @@ -870,16 +870,16 @@
8.4.2 Spatial co-expression modul
- Rank spatial correlated clusters and show genes for selected clusters
-
+
Plot the correlation and number of spatial genes in each cluster.
-top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- selected_clusters = 6,
- show_top_feats = 1)
+
+
8.4.2 Spatial co-expression modul
-
top_netw_spat_cluster <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ selected_clusters = 6,
+ show_top_feats = 1)Figure 8.14: Ranking of spatial correlated groups. Size indicates the number spatial genes per group. @@ -888,23 +888,23 @@
8.4.2 Spatial co-expression modul
- Create the metagene enrichment score per co-expression cluster
-cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")
+cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")
cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
- use_clus_name = "spat_netw_clus",
- show_top_feats = 1)
-
-cluster_genes <- cluster_genes_DT$clus
-names(cluster_genes) <- cluster_genes_DT$feat_ID
-
-visium_brain <- createMetafeats(visium_brain,
- feat_clusters = cluster_genes,
- name = "cluster_metagene")cluster_genes_DT <- showSpatialCorFeats(spat_cor_netw_DT,
+ use_clus_name = "spat_netw_clus",
+ show_top_feats = 1)
+
+cluster_genes <- cluster_genes_DT$clus
+names(cluster_genes) <- cluster_genes_DT$feat_ID
+
+visium_brain <- createMetafeats(visium_brain,
+ feat_clusters = cluster_genes,
+ name = "cluster_metagene")Plot the spatial distribution of the metagene enrichment scores of each spatial co-expression cluster.
-spatCellPlot(visium_brain,
- spat_enr_names = "cluster_metagene",
- cell_annotation_values = netw_ranks$clusters,
- point_size = 1,
- cow_n_col = 5)
+
+
8.5 Spatially informed clusters
spatCellPlot(visium_brain,
+ spat_enr_names = "cluster_metagene",
+ cell_annotation_values = netw_ranks$clusters,
+ point_size = 1,
+ cow_n_col = 5)Figure 8.15: Spatial distribution of metagene enrichment scores per co-expression cluster. @@ -917,56 +917,56 @@
8.5 Spatially informed clusters
Get the top 30 genes per spatial co-expression cluster
-coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genes
+coexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genes
- Re-calculate the clustering
Use the spatial genes to calculate again the principal components, umap, network and clustering
-visium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")
+visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")
- Visualize
Plot the spatial distribution of the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
-
-
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes.
@@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
+HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")
Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")
+visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")
- Visualize
Plot the spatial distribution of the HMRF domains.
-
-
+
+
Figure 8.18: Spatial distribution of HMRF domains.
@@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
+brain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot
- Run the Shiny app
-
-
+
+
Figure 8.19: Shiny app using the visium brain sample.
@@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
-
+
+
Figure 8.20: Polygons selected using the interactive Shiny app.
@@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
+giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)
- Add the polygons to the Giotto object
-
+
- Add the corresponding polygon IDs to the cell metadata
-
+
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
-
+
If no polygon name is provided, the function will retrieve cells located within all polygons
-
+
- Compare the expression levels of some genes of interest between the selected regions
-
-
+
+
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon.
@@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
-
+scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)
+
Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon.
@@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
-
+
+
Figure 8.23: Heatmap showing the cell abundance per selected polygon.
@@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
-
+
+
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon.
@@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
-
+spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)
+
Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons.
@@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
-
+spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)
+
Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons.
@@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
-
+
+
Figure 8.27: Spatial location of selected polygons.
@@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@ 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
coexpr_dt <- data.table::data.table(
- genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
- cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
-
-data.table::setorder(coexpr_dt, cluster)
-
-top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
-
-spatial_genes <- top30_coexpr_dt$genescoexpr_dt <- data.table::data.table(
+ genes = names(spat_cor_netw_DT$cor_clusters$spat_netw_clus),
+ cluster = spat_cor_netw_DT$cor_clusters$spat_netw_clus)
+
+data.table::setorder(coexpr_dt, cluster)
+
+top30_coexpr_dt <- coexpr_dt[, head(.SD, 30) , by = cluster]
+
+spatial_genes <- top30_coexpr_dt$genesvisium_brain <- runPCA(gobject = visium_brain,
- feats_to_use = spatial_genes,
- name = "custom_pca")
-
-visium_brain <- runUMAP(visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- name = "custom_umap")
-
-visium_brain <- createNearestNetwork(gobject = visium_brain,
- dim_reduction_name = "custom_pca",
- dimensions_to_use = 1:20,
- k = 5,
- name = "custom_NN")
-
-visium_brain <- doLeidenCluster(gobject = visium_brain,
- network_name = "custom_NN",
- resolution = 0.15,
- n_iterations = 1000,
- name = "custom_leiden")visium_brain <- runPCA(gobject = visium_brain,
+ feats_to_use = spatial_genes,
+ name = "custom_pca")
+
+visium_brain <- runUMAP(visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ name = "custom_umap")
+
+visium_brain <- createNearestNetwork(gobject = visium_brain,
+ dim_reduction_name = "custom_pca",
+ dimensions_to_use = 1:20,
+ k = 5,
+ name = "custom_NN")
+
+visium_brain <- doLeidenCluster(gobject = visium_brain,
+ network_name = "custom_NN",
+ resolution = 0.15,
+ n_iterations = 1000,
+ name = "custom_leiden")Figure 8.16: Spatial distribution of Leiden clusters calculated using spatial genes.
Plot the UMAP and color the spots using the Leiden clusters calculated based on the spatial genes.
- -
+
+
8.6 Spatial domains HMRF
+
+
Figure 8.17: UMAP plot, colors indicate the Leiden clusters calculated using spatial genes. @@ -980,26 +980,26 @@
8.6 Spatial domains HMRFHMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")
HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
- expression_values = "scaled",
- spatial_genes = spatial_genes,
- k = 20,
- spatial_network_name = "spatial_network",
- betas = c(0, 10, 5),
- output_folder = "11_HMRF/")HMRF_spatial_genes <- doHMRF(gobject = visium_brain,
+ expression_values = "scaled",
+ spatial_genes = spatial_genes,
+ k = 20,
+ spatial_network_name = "spatial_network",
+ betas = c(0, 10, 5),
+ output_folder = "11_HMRF/")Add the HMRF results to the giotto object
-visium_brain <- addHMRF(gobject = visium_brain,
- HMRFoutput = HMRF_spatial_genes,
- k = 20,
- betas_to_add = c(0, 10, 20, 30, 40),
- hmrf_name = "HMRF")visium_brain <- addHMRF(gobject = visium_brain,
+ HMRFoutput = HMRF_spatial_genes,
+ k = 20,
+ betas_to_add = c(0, 10, 20, 30, 40),
+ hmrf_name = "HMRF")- Visualize
Plot the spatial distribution of the HMRF domains.
- -
+
+
8.7 Interactive tools
+
Figure 8.18: Spatial distribution of HMRF domains. @@ -1012,18 +1012,18 @@
8.7 Interactive toolsbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlot
brain_spatPlot <- spatPlot2D(gobject = visium_brain,
- cell_color = "leiden_clus",
- show_image = FALSE,
- return_plot = TRUE,
- point_size = 1)
-
-brain_spatPlotbrain_spatPlot <- spatPlot2D(gobject = visium_brain,
+ cell_color = "leiden_clus",
+ show_image = FALSE,
+ return_plot = TRUE,
+ point_size = 1)
+
+brain_spatPlot- Run the Shiny app
+
+
8.7 Interactive tools
-
Figure 8.19: Shiny app using the visium brain sample. @@ -1032,8 +1032,8 @@
8.7 Interactive toolspolygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
polygon_coordinates <- plotInteractivePolygons(brain_spatPlot)
+
+
8.7 Interactive tools
+
Figure 8.20: Polygons selected using the interactive Shiny app. @@ -1042,34 +1042,34 @@
8.7 Interactive toolsgiotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)
giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
- name = "selections",
- calc_centroids = TRUE)giotto_polygons <- createGiottoPolygonsFromDfr(polygon_coordinates,
+ name = "selections",
+ calc_centroids = TRUE)- Add the polygons to the Giotto object
- Add the corresponding polygon IDs to the cell metadata
- Extract the coordinates and IDs from cells located within one or multiple regions of interest.
If no polygon name is provided, the function will retrieve cells located within all polygons
- +- Compare the expression levels of some genes of interest between the selected regions
+
+
8.7 Interactive tools
-
Figure 8.21: Heatmap showing the z-scores of three genes per selected polygon. @@ -1078,20 +1078,20 @@
8.7 Interactive toolsscran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
scran_results <- findMarkers_one_vs_all(
- visium_brain,
- spat_unit = "cell",
- feat_type = "rna",
- method = "scran",
- expression_values = "normalized",
- cluster_column = "selections",
- min_feats = 2)
-
-top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
-
-comparePolygonExpression(visium_brain,
- selected_feats = top_genes)
+
+
8.7 Interactive tools
-
scran_results <- findMarkers_one_vs_all(
+ visium_brain,
+ spat_unit = "cell",
+ feat_type = "rna",
+ method = "scran",
+ expression_values = "normalized",
+ cluster_column = "selections",
+ min_feats = 2)
+
+top_genes <- scran_results[, head(.SD, 2), by = "cluster"]$feats
+
+comparePolygonExpression(visium_brain,
+ selected_feats = top_genes)Figure 8.22: Heatmap showing the z-scores of top scran genes per selected polygon. @@ -1100,8 +1100,8 @@
8.7 Interactive toolscompareCellAbundance(visium_brain)
compareCellAbundance(visium_brain)
+
+
8.7 Interactive tools
-
Figure 8.23: Heatmap showing the cell abundance per selected polygon. @@ -1110,9 +1110,9 @@
8.7 Interactive toolscompareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
compareCellAbundance(visium_brain,
- cell_type_column = "custom_leiden")
+
+
8.7 Interactive tools
-
Figure 8.24: Heatmap showing the Leiden clusters abundance per selected polygon. @@ -1121,13 +1121,13 @@
8.7 Interactive toolsspatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
spatPlot2D(visium_brain,
- cell_color = "leiden_clus",
- group_by = "selections",
- cow_n_col = 3,
- point_size = 2,
- show_legend = FALSE)
+
+
8.7 Interactive tools
-
spatPlot2D(visium_brain,
+ cell_color = "leiden_clus",
+ group_by = "selections",
+ cow_n_col = 3,
+ point_size = 2,
+ show_legend = FALSE)Figure 8.25: Spatial distribution of Leiden clusters across the selected polygons. @@ -1136,12 +1136,12 @@
8.7 Interactive toolsspatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
spatFeatPlot2D(visium_brain,
- expression_values = "scaled",
- group_by = "selections",
- feats = "Psd",
- point_size = 2)
+
+
8.7 Interactive tools
-12.2.1 Dataset
spatFeatPlot2D(visium_brain,
+ expression_values = "scaled",
+ group_by = "selections",
+ feats = "Psd",
+ point_size = 2)Figure 8.26: Spatial distribution of Psd scaled expression across the selected polygons. @@ -1150,10 +1150,10 @@
8.7 Interactive toolsplotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
plotPolygons(visium_brain,
- polygon_name = "selections",
- x = brain_spatPlot)
+
+
8.7 Interactive tools
diff --git a/working-with-multiple-samples.html b/working-with-multiple-samples.html
index d1a26ed..ce4e7dd 100644
--- a/working-with-multiple-samples.html
+++ b/working-with-multiple-samples.html
@@ -529,7 +529,7 @@
Figure 8.27: Spatial location of selected polygons. @@ -1162,105 +1162,105 @@
8.7 Interactive tools
8.8 Session info
-
-R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8
+
+R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8
R version 4.4.1 (2024-06-14)
-Platform: aarch64-apple-darwin20
-Running under: macOS Sonoma 14.5
-
-Matrix products: default
-BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
-LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
-
-locale:
-[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
-
-time zone: America/New_York
-tzcode source: internal
-
-attached base packages:
-[1] stats graphics grDevices utils datasets methods base
-
-other attached packages:
-[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
-
-loaded via a namespace (and not attached):
- [1] later_1.3.2 tibble_3.2.1
- [3] R.oo_1.26.0 polyclip_1.10-7
- [5] lifecycle_1.0.4 edgeR_4.2.1
- [7] doParallel_1.0.17 lattice_0.22-6
- [9] MASS_7.3-61 backports_1.5.0
- [11] magrittr_2.0.3 sass_0.4.9
- [13] limma_3.60.4 plotly_4.10.4
- [15] rmarkdown_2.27 jquerylib_0.1.4
- [17] yaml_2.3.9 metapod_1.12.0
- [19] httpuv_1.6.15 sp_2.1-4
- [21] reticulate_1.38.0 cowplot_1.1.3
- [23] RColorBrewer_1.1-3 abind_1.4-5
- [25] zlibbioc_1.50.0 quadprog_1.5-8
- [27] GenomicRanges_1.56.1 purrr_1.0.2
- [29] R.utils_2.12.3 BiocGenerics_0.50.0
- [31] tweenr_2.0.3 circlize_0.4.16
- [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
- [35] S4Vectors_0.42.1 ggrepel_0.9.5
- [37] irlba_2.3.5.1 terra_1.7-78
- [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
- [41] colorRamp2_0.1.0 codetools_0.2-20
- [43] DelayedArray_0.30.1 scuttle_1.14.0
- [45] ggforce_0.4.2 tidyselect_1.2.1
- [47] shape_1.4.6.1 UCSC.utils_1.0.0
- [49] farver_2.1.2 ScaledMatrix_1.12.0
- [51] matrixStats_1.3.0 stats4_4.4.1
- [53] GiottoData_0.2.12.0 jsonlite_1.8.8
- [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
- [57] progressr_0.14.0 iterators_1.0.14
- [59] systemfonts_1.1.0 foreach_1.5.2
- [61] dbscan_1.2-0 tools_4.4.1
- [63] ragg_1.3.2 Rcpp_1.0.13
- [65] glue_1.7.0 SparseArray_1.4.8
- [67] xfun_0.46 MatrixGenerics_1.16.0
- [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
- [71] withr_3.0.0 fastmap_1.2.0
- [73] bluster_1.14.0 fansi_1.0.6
- [75] digest_0.6.36 rsvd_1.0.5
- [77] R6_2.5.1 mime_0.12
- [79] textshaping_0.4.0 colorspace_2.1-0
- [81] scattermore_1.2 Cairo_1.6-2
- [83] gtools_3.9.5 R.methodsS3_1.8.2
- [85] utf8_1.2.4 tidyr_1.3.1
- [87] generics_0.1.3 data.table_1.15.4
- [89] FNN_1.1.4 httr_1.4.7
- [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
- [93] scatterpie_0.2.3 uwot_0.2.2
- [95] pkgconfig_2.0.3 gtable_0.3.5
- [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
- [99] SingleCellExperiment_1.26.0 XVector_0.44.0
-[101] htmltools_0.5.8.1 bookdown_0.40
-[103] clue_0.3-65 scales_1.3.0
-[105] Biobase_2.64.0 GiottoUtils_0.1.10
-[107] png_0.1-8 SpatialExperiment_1.14.0
-[109] scran_1.32.0 ggfun_0.1.5
-[111] knitr_1.48 rstudioapi_0.16.0
-[113] reshape2_1.4.4 rjson_0.2.21
-[115] checkmate_2.3.1 cachem_1.1.0
-[117] GlobalOptions_0.1.2 stringr_1.5.1
-[119] parallel_4.4.1 miniUI_0.1.1.1
-[121] RcppZiggurat_0.1.6 pillar_1.9.0
-[123] grid_4.4.1 vctrs_0.6.5
-[125] promises_1.3.0 BiocSingular_1.20.0
-[127] beachmat_2.20.0 xtable_1.8-4
-[129] cluster_2.1.6 evaluate_0.24.0
-[131] magick_2.8.4 cli_3.6.3
-[133] locfit_1.5-9.10 compiler_4.4.1
-[135] rlang_1.1.4 crayon_1.5.3
-[137] labeling_0.4.3 plyr_1.8.9
-[139] stringi_1.8.4 viridisLite_0.4.2
-[141] deldir_2.0-4 BiocParallel_1.38.0
-[143] munsell_0.5.1 lazyeval_0.2.2
-[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
-[147] ggplot2_3.5.1 statmod_1.5.0
-[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
-[151] memoise_2.0.1 igraph_2.0.3
-[153] bslib_0.7.0 RcppParallel_5.1.8 R version 4.4.1 (2024-06-14)
+Platform: aarch64-apple-darwin20
+Running under: macOS Sonoma 14.5
+
+Matrix products: default
+BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
+LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0
+
+locale:
+[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
+
+time zone: America/New_York
+tzcode source: internal
+
+attached base packages:
+[1] stats graphics grDevices utils datasets methods base
+
+other attached packages:
+[1] shiny_1.8.1.1 Giotto_4.1.0 GiottoClass_0.3.3
+
+loaded via a namespace (and not attached):
+ [1] later_1.3.2 tibble_3.2.1
+ [3] R.oo_1.26.0 polyclip_1.10-7
+ [5] lifecycle_1.0.4 edgeR_4.2.1
+ [7] doParallel_1.0.17 lattice_0.22-6
+ [9] MASS_7.3-61 backports_1.5.0
+ [11] magrittr_2.0.3 sass_0.4.9
+ [13] limma_3.60.4 plotly_4.10.4
+ [15] rmarkdown_2.27 jquerylib_0.1.4
+ [17] yaml_2.3.9 metapod_1.12.0
+ [19] httpuv_1.6.15 sp_2.1-4
+ [21] reticulate_1.38.0 cowplot_1.1.3
+ [23] RColorBrewer_1.1-3 abind_1.4-5
+ [25] zlibbioc_1.50.0 quadprog_1.5-8
+ [27] GenomicRanges_1.56.1 purrr_1.0.2
+ [29] R.utils_2.12.3 BiocGenerics_0.50.0
+ [31] tweenr_2.0.3 circlize_0.4.16
+ [33] GenomeInfoDbData_1.2.12 IRanges_2.38.1
+ [35] S4Vectors_0.42.1 ggrepel_0.9.5
+ [37] irlba_2.3.5.1 terra_1.7-78
+ [39] dqrng_0.4.1 DelayedMatrixStats_1.26.0
+ [41] colorRamp2_0.1.0 codetools_0.2-20
+ [43] DelayedArray_0.30.1 scuttle_1.14.0
+ [45] ggforce_0.4.2 tidyselect_1.2.1
+ [47] shape_1.4.6.1 UCSC.utils_1.0.0
+ [49] farver_2.1.2 ScaledMatrix_1.12.0
+ [51] matrixStats_1.3.0 stats4_4.4.1
+ [53] GiottoData_0.2.12.0 jsonlite_1.8.8
+ [55] GetoptLong_1.0.5 BiocNeighbors_1.22.0
+ [57] progressr_0.14.0 iterators_1.0.14
+ [59] systemfonts_1.1.0 foreach_1.5.2
+ [61] dbscan_1.2-0 tools_4.4.1
+ [63] ragg_1.3.2 Rcpp_1.0.13
+ [65] glue_1.7.0 SparseArray_1.4.8
+ [67] xfun_0.46 MatrixGenerics_1.16.0
+ [69] GenomeInfoDb_1.40.1 dplyr_1.1.4
+ [71] withr_3.0.0 fastmap_1.2.0
+ [73] bluster_1.14.0 fansi_1.0.6
+ [75] digest_0.6.36 rsvd_1.0.5
+ [77] R6_2.5.1 mime_0.12
+ [79] textshaping_0.4.0 colorspace_2.1-0
+ [81] scattermore_1.2 Cairo_1.6-2
+ [83] gtools_3.9.5 R.methodsS3_1.8.2
+ [85] utf8_1.2.4 tidyr_1.3.1
+ [87] generics_0.1.3 data.table_1.15.4
+ [89] FNN_1.1.4 httr_1.4.7
+ [91] htmlwidgets_1.6.4 S4Arrays_1.4.1
+ [93] scatterpie_0.2.3 uwot_0.2.2
+ [95] pkgconfig_2.0.3 gtable_0.3.5
+ [97] ComplexHeatmap_2.20.0 GiottoVisuals_0.2.4
+ [99] SingleCellExperiment_1.26.0 XVector_0.44.0
+[101] htmltools_0.5.8.1 bookdown_0.40
+[103] clue_0.3-65 scales_1.3.0
+[105] Biobase_2.64.0 GiottoUtils_0.1.10
+[107] png_0.1-8 SpatialExperiment_1.14.0
+[109] scran_1.32.0 ggfun_0.1.5
+[111] knitr_1.48 rstudioapi_0.16.0
+[113] reshape2_1.4.4 rjson_0.2.21
+[115] checkmate_2.3.1 cachem_1.1.0
+[117] GlobalOptions_0.1.2 stringr_1.5.1
+[119] parallel_4.4.1 miniUI_0.1.1.1
+[121] RcppZiggurat_0.1.6 pillar_1.9.0
+[123] grid_4.4.1 vctrs_0.6.5
+[125] promises_1.3.0 BiocSingular_1.20.0
+[127] beachmat_2.20.0 xtable_1.8-4
+[129] cluster_2.1.6 evaluate_0.24.0
+[131] magick_2.8.4 cli_3.6.3
+[133] locfit_1.5-9.10 compiler_4.4.1
+[135] rlang_1.1.4 crayon_1.5.3
+[137] labeling_0.4.3 plyr_1.8.9
+[139] stringi_1.8.4 viridisLite_0.4.2
+[141] deldir_2.0-4 BiocParallel_1.38.0
+[143] munsell_0.5.1 lazyeval_0.2.2
+[145] Matrix_1.7-0 sparseMatrixStats_1.16.0
+[147] ggplot2_3.5.1 statmod_1.5.0
+[149] SummarizedExperiment_1.34.0 Rfast_2.1.0
+[151] memoise_2.0.1 igraph_2.0.3
+[153] bslib_0.7.0 RcppParallel_5.1.8 12.2.1 Dataset
12.2.2 Visium technology
-
+
Figure 12.1: Overview of Visium. Source: 10X Genomics.
@@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
Figure 12.1: Overview of Visium. Source: 10X Genomics. @@ -543,67 +543,67 @@
12.3 Create individual giotto obj
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
+data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))
12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
-
+
12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
+print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
Figure 12.2: Structure of Giotto object containing a single dataset.
@@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
12.3.1 Download the data
You need to download the expression matrix and spatial information by running these commands:
-data_dir <- "data/3_session1"
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- showWarnings = TRUE, recursive = TRUE)
-
-# Spatial data adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
-
-# Download matrix adenocarcinoma prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
-
-dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- showWarnings = FALSE, recursive = TRUE)
-
-# Spatial data normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
-
-# Download matrix normal prostate
-download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
- destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))data_dir <- "data/3_session1"
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ showWarnings = TRUE, recursive = TRUE)
+
+# Spatial data adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_spatial.tar.gz"))
+
+# Download matrix adenocarcinoma prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer/Visium_FFPE_Human_Prostate_Cancer_raw_feature_bc_matrix.tar.gz"))
+
+dir.create(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ showWarnings = FALSE, recursive = TRUE)
+
+# Spatial data normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_spatial.tar.gz"))
+
+# Download matrix normal prostate
+download.file(url = "https://cf.10xgenomics.com/samples/spatial-exp/1.3.0/Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz",
+ destfile = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate/Visium_FFPE_Human_Normal_Prostate_raw_feature_bc_matrix.tar.gz"))12.3.2 Create giotto instructions
We must first create instructions for our Giotto object. This will tell the object where to save outputs, whether to show or return plots, and the python path. Specifying the python path is often not required as Giotto will identify the relevant python environment, but might be required in some instances.
- +12.3.3 Load visium data into Giotto
We next need to read in the data for the Giotto object. To do this we will use the createGiottoVisiumObject() convenience function. This requires us to specify the directory that contains the visium data output from 10X Genomics’s Spaceranger. We also specify the expression data to use (raw or filtered) as well as the image to align. Spaceranger outputs two images, a low and high resolution image.
-print(data_dir)
-print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
-
-## Healthy prostate
-N_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)
-
-## Adenocarcinoma
-C_pros <- createGiottoVisiumObject(
- visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
- expr_data = "raw",
- png_name = "tissue_lowres_image.png",
- gene_column_index = 2,
- instructions = instrs
-)print(data_dir)
+print(file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"))
+
+## Healthy prostate
+N_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Normal_Prostate"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)
+
+## Adenocarcinoma
+C_pros <- createGiottoVisiumObject(
+ visium_dir = file.path(data_dir, "Visium_FFPE_Human_Prostate_Cancer"),
+ expr_data = "raw",
+ png_name = "tissue_lowres_image.png",
+ gene_column_index = 2,
+ instructions = instrs
+)We can see that the gobject contains information for the cells (polygon and spatial units), the RNA express (raw) and the relevant image.
-
+
12.3.3 Load visium data into Giot
Figure 12.2: Structure of Giotto object containing a single dataset. @@ -613,14 +613,14 @@
12.3.3 Load visium data into Giot
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
-
+spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
Figure 12.3: Tissue coverage for the normal prostate sample.
@@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
-
+spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample.
@@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)
+combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)
From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
12.3.4 Healthy prostate tissue coverage
Aligning the Visium spots to the tissue using the fiducials that border the capture area enables the identification of spots containing expression data from the tissue. These spots can be visualized using the spatPlot2D function by setting the cell_color parameter to “in_tissue”.
-spatPlot2D(gobject = N_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_normal_prostate_tissue"))
+
+
-
+
spatPlot2D(gobject = N_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_normal_prostate_tissue"))Figure 12.3: Tissue coverage for the normal prostate sample. @@ -629,14 +629,14 @@
12.3.4 Healthy prostate tissue co
12.3.5 Adenocarcinoma prostate tissue coverage
-spatPlot2D(gobject = C_pros,
- cell_color = "in_tissue",
- show_image = TRUE,
- point_size = 2.5,
- cell_color_code = c("black", "red"),
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))
+
+
spatPlot2D(gobject = C_pros,
+ cell_color = "in_tissue",
+ show_image = TRUE,
+ point_size = 2.5,
+ cell_color_code = c("black", "red"),
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_adeno_prostate_tissue"))Figure 12.4: Tissue coverage for the adenocarcinoma prostate sample. @@ -645,23 +645,23 @@
12.3.5 Adenocarcinoma prostate ti
12.4 Join Giotto Objects
To join objects together we can use the joinGittoObjects() function. For this we need to supply a list of objects as well as the names for each of these objects. We can also specify the x and y padding to separate the objects in space or the Z position for 3D datasets. If the x_shift is set to NULL then the total shift will be guessed from the Giotto image.
-combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
- gobject_names = c("NP", "CP"),
- join_method = "shift", x_padding = 1000)
-
-# Printing the file structure for the individual datasets
-print(head(pDataDT(combined_pros)))
-print(combined_pros)combined_pros <- joinGiottoObjects(gobject_list = list(N_pros, C_pros),
+ gobject_names = c("NP", "CP"),
+ join_method = "shift", x_padding = 1000)
+
+# Printing the file structure for the individual datasets
+print(head(pDataDT(combined_pros)))
+print(combined_pros)From the joined data we can see the same information that was present in the single dataset objects as well as the addition of another image. The images are renamed from “image” to include the object name in the image name e.g. “NP-image”. We can also see in the cell metadata that there is a new column “list_ID” that contains the original object names. The cell_ID column also has the original object name appended to the beginning of each cell ID e.g. “NP-AAACAACGAATAGTTC-1”.
-
+
12.5 Visualizing combined dataset
Figure 12.5: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. @@ -674,15 +674,15 @@
12.5 Visualizing combined dataset
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))
+
Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot.
@@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
12.5.1 Vizualizing in the same plot
Due to the x_padding provided when joining the objects each of the datasets can be visualized in the same plotting area. We can see below the normal prostate sample on the left and the healthy prostate on the right. By including the show_image function and supplying both of the image names (“NP-image”, “CP-image”), we can also include the relevant images within the same plot.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- point_size = 1,
- point_alpha = 0.5,
- save_param = list(save_name = "03_ses1_combined_tissue"))
+
+
12.5.1 Vizualizing in the same pl
spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ point_size = 1,
+ point_alpha = 0.5,
+ save_param = list(save_name = "03_ses1_combined_tissue"))Figure 12.6: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) within the same plot. @@ -692,17 +692,17 @@
12.5.1 Vizualizing in the same pl
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
-
+spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots.
@@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
12.5.2 Vizualizing on separate plots
If we want to visualize the datasets in separate plots we can supply the “group_by” variable. Below we group the data by “list_ID”, which corresponds to each dataset. We can specify the number of columns through the “cow_n_col” variable.
-spatPlot2D(gobject = combined_pros,
- cell_color = "in_tissue",
- cell_color_code = c("black", "pink"),
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- group_by = "list_ID",
- point_alpha = 0.5,
- point_size = 0.5,
- cow_n_col = 1,
- save_param = list(save_name = "03_ses1_combined_tissue_group"))
+
+
12.5.2 Vizualizing on separate pl
spatPlot2D(gobject = combined_pros,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "pink"),
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ group_by = "list_ID",
+ point_alpha = 0.5,
+ point_size = 0.5,
+ cow_n_col = 1,
+ save_param = list(save_name = "03_ses1_combined_tissue_group"))Figure 12.7: Vizualizing the visium spots that overlap tissue in normal prostate (left) and adenocarcinoma samples (right) in separate plots. @@ -713,23 +713,23 @@
12.5.2 Vizualizing on separate pl
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
-
+# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))
+
Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset.
@@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
12.6 Splitting combined dataset
If needed it’s possible to split the individual objects into single objects again through subsetting the cell metadata as shown below.
-# Getting the cell information
-combined_cells <- pDataDT(combined_pros)
-np_cells <- combined_cells[list_ID == "NP"]
-
-np_split <- subsetGiotto(combined_pros,
- cell_ids = np_cells$cell_ID,
- poly_info = np_cells$cell_ID,
- spat_unit = "cell")
-
-spatPlot2D(gobject = np_split,
- cell_color = "in_tissue",
- cell_color_code = c("black", "red"),
- show_image = TRUE,
- point_alpha = 0.5,
- point_size = 0.5,
- save_param = list(save_name = "03_ses1_split_object"))
+
+
12.7 Analyzing joined objects
# Getting the cell information
+combined_cells <- pDataDT(combined_pros)
+np_cells <- combined_cells[list_ID == "NP"]
+
+np_split <- subsetGiotto(combined_pros,
+ cell_ids = np_cells$cell_ID,
+ poly_info = np_cells$cell_ID,
+ spat_unit = "cell")
+
+spatPlot2D(gobject = np_split,
+ cell_color = "in_tissue",
+ cell_color_code = c("black", "red"),
+ show_image = TRUE,
+ point_alpha = 0.5,
+ point_size = 0.5,
+ save_param = list(save_name = "03_ses1_split_object"))Figure 12.8: Structure of Giotto object containing two datasets (left) and cell metadata on the left. Note the addition of multiple images and the addition of the list_ID column to define the dataset. @@ -741,38 +741,38 @@
12.7 Analyzing joined objects
12.7.1 Normalization and adding statistics
Now that the objects have been joined we can analyze the object as if it was a single object. This means all of the analyses will be performed in parallel. Therefore, all of the filtering and normalization will be identical between datasets.
-# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))
+# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))
After performing the addStatistics() function on both the datasets we can see the relative expression for each spot in both samples.
-
+
Figure 12.9: Unique feat expression for visium spots for both prostate samples.
@@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
# subset on in-tissue spots
-metadata <- pDataDT(combined_pros)
-in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
-combined_pros <- subsetGiotto(combined_pros,
- cell_ids = in_tissue_barcodes)
-
-## filter
-combined_pros <- filterGiotto(gobject = combined_pros,
- expression_threshold = 1,
- feat_det_in_min_cells = 50,
- min_det_feats_per_cell = 500,
- expression_values = "raw",
- verbose = TRUE)
-
-## normalize
-combined_pros <- normalizeGiotto(gobject = combined_pros,
- scalefactor = 6000)
-
-## add gene & cell statistics
-combined_pros <- addStatistics(gobject = combined_pros,
- expression_values = "raw")
-
-## visualize
-spatPlot2D(gobject = combined_pros,
- cell_color = "nr_feats",
- color_as_factor = FALSE,
- point_size = 1,
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_feat_expression"))# subset on in-tissue spots
+metadata <- pDataDT(combined_pros)
+in_tissue_barcodes <- metadata[in_tissue == 1]$cell_ID
+combined_pros <- subsetGiotto(combined_pros,
+ cell_ids = in_tissue_barcodes)
+
+## filter
+combined_pros <- filterGiotto(gobject = combined_pros,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 50,
+ min_det_feats_per_cell = 500,
+ expression_values = "raw",
+ verbose = TRUE)
+
+## normalize
+combined_pros <- normalizeGiotto(gobject = combined_pros,
+ scalefactor = 6000)
+
+## add gene & cell statistics
+combined_pros <- addStatistics(gobject = combined_pros,
+ expression_values = "raw")
+
+## visualize
+spatPlot2D(gobject = combined_pros,
+ cell_color = "nr_feats",
+ color_as_factor = FALSE,
+ point_size = 1,
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_feat_expression"))Figure 12.9: Unique feat expression for visium spots for both prostate samples. @@ -782,38 +782,38 @@
12.7.1 Normalization and adding s
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)
+## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)
12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))
+
Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
12.7.2 Clustering the datasets
Since we shifted the objects within space the spatial networks for each dataset will remain separate, assuming that the lower limits for neighbors is smaller than the distance of each dataset. However, the individual spot clustering will be performed on all spots from both datasets as if they were a single object, meaning that the same cell types between objects should be clustered together
-## PCA ##
-combined_pros <- calculateHVF(gobject = combined_pros)
-
-combined_pros <- runPCA(gobject = combined_pros,
- center = TRUE,
- scale_unit = TRUE)
-
-## cluster and run UMAP ##
-# sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "pca",
- dim_reduction_name = "pca",
- dimensions_to_use = 1:10,
- k = 15)
-
-# Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- resolution = 0.2,
- n_iterations = 200)
-
-# UMAP
-combined_pros <- runUMAP(combined_pros)## PCA ##
+combined_pros <- calculateHVF(gobject = combined_pros)
+
+combined_pros <- runPCA(gobject = combined_pros,
+ center = TRUE,
+ scale_unit = TRUE)
+
+## cluster and run UMAP ##
+# sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "pca",
+ dim_reduction_name = "pca",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+# Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ resolution = 0.2,
+ n_iterations = 200)
+
+# UMAP
+combined_pros <- runUMAP(combined_pros)12.7.3 Vizualizing spatial location of clusters
We can visualize the clusters determined through Leiden clustering on both of the datasets within the same plot.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "leiden_clus",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_leiden_clus"))
+
+
12.7.3 Vizualizing spatial locati
spatDimPlot2D(gobject = combined_pros,
+ cell_color = "leiden_clus",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_leiden_clus"))Figure 12.10: UMAP (top) for both samples colored by Leiden clusters visualized in a spatial plot (bottom) for the normal prostate (left) and the adenocarcinoma prostate sample (right). @@ -823,12 +823,12 @@
12.7.3 Vizualizing spatial locati
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))
+
Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")
+library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")
After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
Figure 12.12: Data structure of the gobject after running Harmony integration.
@@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
12.7.4 Vizualizing tissue contribution to clusters
We can also color the UMAP to visualize the contribution from each tissue in the UMAP. To do this we color the UMAP by “list_ID” rather than “leiden_clus”. If each of the cell types between both samples cluster together then we would expect that clusters should contain the cell color of both samples. However, we can see that the samples are clustered distinctly within the UMAP. This indicates that the cell types shared between both samples are found within different clusters indicating that more complex integration techniques might be required for these samples.
-spatDimPlot2D(gobject = combined_pros,
- cell_color = "list_ID",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- save_param = list(save_name = "ses3_1_tissue_contribution"))
+
+
12.7.4 Vizualizing tissue contrib
+
spatDimPlot2D(gobject = combined_pros,
+ cell_color = "list_ID",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ save_param = list(save_name = "ses3_1_tissue_contribution"))Figure 12.11: Tissue contribution for leiden clustering for the normal prostate (left) and the adenocarcinoma prostate sample (right). @@ -840,13 +840,13 @@
12.7.4 Vizualizing tissue contrib
12.8 Perform Harmony and default workflows
We can use Harmony to integrate multiple datasets, grouping equivelent cell types between samples. Harmony is an algorithm that iteratively adjusts cell coordinates in a reduced-dimensional space to correct for dataset-specific effects. It uses fuzzy clustering to assign cells to multiple clusters, calculates dataset-specific correction factors, and applies these corrections to each cell, repeating the process until the influence of the dataset diminishes.
Before running Harmony we need to run the PCA function or set “do_pca” to TRUE. We ran this above so do not need to perform this step. Harmony will default to attempting 10 rounds of integration. Not all samples will need the full 10 and will finish accordingly. The following dataset should converge after 5 iterations.
-library(harmony)
-
-## run harmony integration
-combined_pros <- runGiottoHarmony(combined_pros,
- vars_use = "list_ID")library(harmony)
+
+## run harmony integration
+combined_pros <- runGiottoHarmony(combined_pros,
+ vars_use = "list_ID")After running the Harmony function successfully we can see that the outputted gobject has a new dim reduction names “harmony”. We can use this for all subsequent spatial steps.
-
+
12.8 Perform Harmony and default
Figure 12.12: Data structure of the gobject after running Harmony integration. @@ -855,37 +855,37 @@
12.8 Perform Harmony and default
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))
+## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))
We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right).
@@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
12.8.1 Clustering harmonized object
We can now perform the same clustering steps as before but instead using the “harmony” dim reduction rather than PCA. We will also be creating new UMAP and nearest network data for the gobject that will be named differently to before to preserve the original analyses. If using the same name then this will overwrite the original analysis.
-## sNN network (default)
-combined_pros <- createNearestNetwork(gobject = combined_pros,
- dim_reduction_to_use = "harmony",
- dim_reduction_name = "harmony",
- name = "NN.harmony",
- dimensions_to_use = 1:10,
- k = 15)
-
-## Leiden clustering
-combined_pros <- doLeidenCluster(gobject = combined_pros,
- network_name = "NN.harmony",
- resolution = 0.2,
- n_iterations = 1000,
- name = "leiden_harmony")
-
-# UMAP dimension reduction
-combined_pros <- runUMAP(combined_pros,
- dim_reduction_name = "harmony",
- dim_reduction_to_use = "harmony",
- name = "umap_harmony")
-
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "leiden_harmony",
- show_image = TRUE,
- image_name = c("NP-image", "CP-image"),
- spat_point_size = 1,
- save_param = list(save_name = "leiden_clustering_harmony"))## sNN network (default)
+combined_pros <- createNearestNetwork(gobject = combined_pros,
+ dim_reduction_to_use = "harmony",
+ dim_reduction_name = "harmony",
+ name = "NN.harmony",
+ dimensions_to_use = 1:10,
+ k = 15)
+
+## Leiden clustering
+combined_pros <- doLeidenCluster(gobject = combined_pros,
+ network_name = "NN.harmony",
+ resolution = 0.2,
+ n_iterations = 1000,
+ name = "leiden_harmony")
+
+# UMAP dimension reduction
+combined_pros <- runUMAP(combined_pros,
+ dim_reduction_name = "harmony",
+ dim_reduction_to_use = "harmony",
+ name = "umap_harmony")
+
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "leiden_harmony",
+ show_image = TRUE,
+ image_name = c("NP-image", "CP-image"),
+ spat_point_size = 1,
+ save_param = list(save_name = "leiden_clustering_harmony"))We can see a different UMAP and clustering to that seen in the original steps above. We can again map these onto the tissue spots and see where the clusters are spatially.
-
+
12.8.1 Clustering harmonized obje
Figure 12.13: Leiden clustering after harmony was performed for the normal prostate (left) and the adenocarcinoma prostate sample (right). @@ -895,13 +895,13 @@
12.8.1 Clustering harmonized obje
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
-
+spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right).
diff --git a/xenium-1.html b/xenium-1.html
index 64e2f5f..1140f26 100644
--- a/xenium-1.html
+++ b/xenium-1.html
@@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)
+library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)
In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
-
+
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset
@@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )
+# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )
ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
-
+
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
@@ -630,11 +630,11 @@ 10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
12.8.2 Vizualizing the tissue contribution
We can see that after performing harmony that the clusters from the two tissue samples are now clustered together. There is still a cluster that is unique to the adenocarcinoma sample, however this is expected as this represents the visium spots that cover the tumor regions of the tissue, which are not found in the normal tissue.
-spatDimPlot2D(gobject = combined_pros,
- dim_reduction_to_use = "umap",
- dim_reduction_name = "umap_harmony",
- cell_color = "list_ID",
- save_plot = TRUE,
- save_param = list(save_name = "leiden_clustering_harmony_contribution"))
+
+
+
10.3 Convenience functionfeature metadata (gene_panel.json)
spatDimPlot2D(gobject = combined_pros,
+ dim_reduction_to_use = "umap",
+ dim_reduction_name = "umap_harmony",
+ cell_color = "list_ID",
+ save_plot = TRUE,
+ save_param = list(save_name = "leiden_clustering_harmony_contribution"))Figure 12.14: Tissue contribution for leiden clustering after harmony for the normal prostate (left) and the adenocarcinoma prostate sample (right). diff --git a/xenium-1.html b/xenium-1.html index 64e2f5f..1140f26 100644 --- a/xenium-1.html +++ b/xenium-1.html @@ -576,24 +576,24 @@
10.1.1 Output directory structure
10.1.2 Mini Xenium Dataset
-library(Giotto)
-# set up paths
-data_path <- "data/02_session3/"
-save_dir <- "results/02_session3/"
-dir.create(save_dir, recursive = TRUE)
-
-# download the mini dataset and untar
-options("timeout" = Inf)
-download.file(
- url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
- destfile = file.path(save_dir, "workshop_xenium.zip")
-)
-untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
- exdir = data_path)library(Giotto)
+# set up paths
+data_path <- "data/02_session3/"
+save_dir <- "results/02_session3/"
+dir.create(save_dir, recursive = TRUE)
+
+# download the mini dataset and untar
+options("timeout" = Inf)
+download.file(
+ url = "https://zenodo.org/records/13207308/files/workshop_xenium.zip?download=1",
+ destfile = file.path(save_dir, "workshop_xenium.zip")
+)
+untar(tarfile = file.path(save_dir, "workshop_xenium.zip"),
+ exdir = data_path)In order to speed up the steps of the workshop and make it locally runnable, we provide a subset of the full dataset.
- Full: -16.039, 12342.984, -3511.515, -294.455 (xmin, xmax, ymin, ymax)
- Mini: 6000, 7000, -2200, -1400 (xmin, xmax, ymin, ymax)
+
+
@@ -630,11 +630,11 @@
Figure 10.1: Shown is the H&E aligned to the Xenium dataset with micron scaling. The blue bounds mark out the area provided as a mini dataset @@ -608,15 +608,15 @@
10.2.1 Image conversion (may chan
First is actually dealing with the image formats. Xenium generates ome.tif images which Giotto is currently not fully compatible with. So we convert them to normal tif images using ometif_to_tif() which works through the python tifffile package.
The image files can then be loaded in downstream steps.
These commented out steps are not needed for today since the mini dataset provides .tif images that have already been spatially aligned and converted. However, the code needed to do this is provided below.
-# image_paths <- list.files(
-# data_path, pattern = "morphology_focus|he_image.ome",
-# recursive = TRUE, full.names = TRUE
-# )# image_paths <- list.files(
+# data_path, pattern = "morphology_focus|he_image.ome",
+# recursive = TRUE, full.names = TRUE
+# )ometif_to_tif() output_dir can be specified, but by default, it writes to a new subdirectory called tif_exports underneath the source image’s directory.
Keep in mind that where the exported tifs get exported to should be where downstream image reading functions should point to. The code run today is with the filepaths that the mini dataset has.
We are also working on a method of directly accessing the ome.tifs for better compatibility in the future.
10.3 Convenience functionfeature metadata (gene_panel.json)
For the full dataset (HPC): time: 1-2min | memory: 24GBC
-?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE
+?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE
There are a lot of other params for additional or alternative items you can load. The next subsections will explain a couple of them.
10.3.1 Specific filepaths
@@ -699,19 +699,19 @@ 10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
?createGiottoXeniumObject
-g <- createGiottoXeniumObject(xenium_dir = data_path)
-# set instructions
-instructions(g, "save_dir") <- save_dir
-instructions(g, "save_plot") <- TRUE?createGiottoXeniumObject
+g <- createGiottoXeniumObject(xenium_dir = data_path)
+# set instructions
+instructions(g, "save_dir") <- save_dir
+instructions(g, "save_plot") <- TRUE10.3.1 Specific filepaths
@@ -699,19 +699,19 @@10.3.3 Transcript type splitting<
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
-
+
10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)
+
Figure 10.2: Simple subcellular plotting to check data
@@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
10.3.4 Centroids calculation
Several Giotto operations require that a set of centroids are calculated for polygon spatial units.
- +10.3.5 Simple visualization
-spatInSituPlotPoints(g,
- polygon_feat_type = "cell",
- feats = list(rna = head(featIDs(g))),
- use_overlap = FALSE,
- polygon_color = "cyan",
- polygon_line_size = 0.1
-)
+
+
10.3.5 Simple visualization
spatInSituPlotPoints(g,
+ polygon_feat_type = "cell",
+ feats = list(rna = head(featIDs(g))),
+ use_overlap = FALSE,
+ polygon_color = "cyan",
+ polygon_line_size = 0.1
+)Figure 10.2: Simple subcellular plotting to check data @@ -722,8 +722,8 @@
10.3.5 Simple visualization
10.4 Piecewise loading
Giotto also provides the importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.
-
+
Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)
importXenium() import utility that allows independent creation of compatible Giotto subobjects for more flexibility.Giotto <XeniumReader>
dir : data/02_session3/
qv_cutoff : 20
@@ -741,17 +741,17 @@ 10.4 Piecewise loading
10.4.1 load giottoPoints transcripts
-
-
+
+
Figure 10.3: plot of Gene expression (‘rna’) density
-
+
An object of class giottoPoints
feat_type : "rna"
Feature Information:
@@ -779,13 +779,13 @@ 10.4.2 (optional) Loading pre-agg
The feat_type of the loaded expression information should be matched to the used feat_type params passed to the convenience function.
The qv_threshold used must be 20 since the 10X outputs are based on that cutoff.
-x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
-ex <- x$load_expression()
-featType(ex)
+x$filetype$expression <- "mtx" # change to mtx instead of .h5 which is not in the mini dataset
+ex <- x$load_expression()
+featType(ex)
[1] "rna" "Negative Control Probe" "Negative Control Codeword"
[4] "Unassigned Codeword"
The feature types here do not match what we established for the transcripts, so we can just change them.
-
+
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -796,19 +796,19 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-featType(ex[[2]]) <- c("NegControlProbe")
-featType(ex[[3]]) <- c("NegControlCodeword")
-featType(ex[[4]]) <- c("UnassignedCodeword")
+featType(ex[[2]]) <- c("NegControlProbe")
+featType(ex[[3]]) <- c("NegControlCodeword")
+featType(ex[[4]]) <- c("UnassignedCodeword")
Then we can just append them to the Giotto object.
Here we set up a second object since we will be using Giotto’s own aggregation method downstream.
-g2 <- g
-# append the expression info
-g2 <- setGiotto(g2, ex)
-
-# load cell metadata
-cx <- x$load_cellmeta()
-g2 <- setGiotto(g2, cx)
-force(g2)
+g2 <- g
+# append the expression info
+g2 <- setGiotto(g2, ex)
+
+# load cell metadata
+cx <- x$load_cellmeta()
+g2 <- setGiotto(g2, cx)
+force(g2)
An object of class giotto
>Active spat_unit: cell
>Active feat_type: rna
@@ -824,32 +824,32 @@ 10.4.2 (optional) Loading pre-agg
spatial locations ----------------
[cell] raw
[nucleus] raw
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "cell_area",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
-spatInSituPlotPoints(g2,
- # polygon shading params
- polygon_fill = "transcript_counts",
- polygon_fill_as_factor = FALSE,
- polygon_fill_gradient_style = "sequential",
- # polygon line params
- polygon_color = "grey",
- polygon_line_size = 0.1
-)
-
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "cell_area",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
+spatInSituPlotPoints(g2,
+ # polygon shading params
+ polygon_fill = "transcript_counts",
+ polygon_fill_as_factor = FALSE,
+ polygon_fill_gradient_style = "sequential",
+ # polygon line params
+ polygon_color = "grey",
+ polygon_line_size = 0.1
+)
+
Figure 10.4: Example plot using 10X metadata. Left is ‘cell_area’, right is ‘transcript_counts’
-
+
@@ -865,13 +865,13 @@ 10.5.1 Image metadataimg_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
-omemeta <- xml2::read_xml(img_xml_path)
-res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
-res <- Reduce(rbind, xml2::xml_attrs(res))
-rownames(res) <- NULL
-res <- as.data.frame(res)
-force(res)
+img_xml_path <- file.path(data_path, "morphology_focus", "morphology_focus_0000.xml")
+omemeta <- xml2::read_xml(img_xml_path)
+res <- xml2::xml_find_all(omemeta, "//d1:Channel", ns = xml2::xml_ns(omemeta))
+res <- Reduce(rbind, xml2::xml_attrs(res))
+rownames(res) <- NULL
+res <- as.data.frame(res)
+force(res)
ID Name SamplesPerPixel
1 Channel:0 DAPI 1
2 Channel:1 18S 1
@@ -881,7 +881,7 @@ 10.5.1 Image metadata
10.5.2 Image loading
morphology_focus images need to be scaled by the micron scaling factor. Aligned images need to first be affine transformed then scaled. The micron scaling factor can be found in the json-like experiment.xenium file under pixel_size (0.2125 for this dataset).
-
+
Figure 10.5: Spatial extent/bounds of transcripts (red), immunofluorescence morphology focus images (blue), H&E aligned image (gold). Lower right shows the affine matrix for aligning the H&E
@@ -912,61 +912,61 @@
10.5.2 Image loadingimg_paths <- c(
- sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
- "data/02_session3/he_mini.tif"
-)
-img_list <- createGiottoLargeImageList(
- img_paths,
- # naming is based on the channel metadata above
- names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
- use_rast_ext = TRUE,
- verbose = FALSE
-)
-
-# make some images brighter
-img_list[[1]]@max_window <- 5000
-img_list[[2]]@max_window <- 5000
-img_list[[3]]@max_window <- 5000
-
-
-# append images to gobject
-g <- setGiotto(g, img_list)
-
-# example plots
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "DAPI",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "18S",
- polygon_feat_type = "cell",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "ATP1A1/CD45/E-Cadherin",
- polygon_feat_type = "nucleus",
- polygon_color = "cyan",
- polygon_line_size = 0.1,
- polygon_alpha = 0
-)
-
+img_paths <- c(
+ sprintf("data/02_session3/morphology_focus/morphology_focus_%04d.tif", 0:3),
+ "data/02_session3/he_mini.tif"
+)
+img_list <- createGiottoLargeImageList(
+ img_paths,
+ # naming is based on the channel metadata above
+ names = c("DAPI", "18S", "ATP1A1/CD45/E-Cadherin", "alphaSMA/Vimentin", "HE"),
+ use_rast_ext = TRUE,
+ verbose = FALSE
+)
+
+# make some images brighter
+img_list[[1]]@max_window <- 5000
+img_list[[2]]@max_window <- 5000
+img_list[[3]]@max_window <- 5000
+
+
+# append images to gobject
+g <- setGiotto(g, img_list)
+
+# example plots
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "DAPI",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "18S",
+ polygon_feat_type = "cell",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "ATP1A1/CD45/E-Cadherin",
+ polygon_feat_type = "nucleus",
+ polygon_color = "cyan",
+ polygon_line_size = 0.1,
+ polygon_alpha = 0
+)
+
Figure 10.6: H&E and Cell polys (top left), DAPI and nuclear polys (top right), 18S and cell polys (lower left), ATP1A1/CD45/E-Cadherin and nuclear polys (lower right)
@@ -977,42 +977,42 @@
10.5.2 Image loading
10.6 Spatial aggregation
First calculate the feat_info ‘rna’ transcripts overlapped by the spatial_info ‘cell’ polygons with calculateOverlap(). Then, the overlaps information (relationships between points and polygons that overlap them) gets converted into a count matrix with overlapToMatrix().
-
+
10.7 Aggregate analyses workflow
10.7.1 Filtering
-# very permissive filtering. Mainly for removing 0 values
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 5
-)
+# very permissive filtering. Mainly for removing 0 values
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 5
+)
Feature type: rna
Number of cells removed: 0 out of 7129
Number of feats removed: 0 out of 377
10.7.2 Normalization
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- polygon_fill = "nr_feats",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-spatInSituPlotPoints(g,
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential",
- polygon_fill_as_factor = FALSE
-)
-
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ polygon_fill = "nr_feats",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential",
+ polygon_fill_as_factor = FALSE
+)
+
Figure 10.7: ‘nr_feats’ - Number of different gene species detected per cell (left), ‘total_expr’ - total detections per cell (right)
@@ -1023,22 +1023,22 @@
10.7.2 Normalization
10.7.3 Dimension Reduction
Dimensional reduction of expression space to visualize expressional differences between cells and help with clustering.
-g <- runPCA(g, feats_to_use = NULL)
-# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
-screePlot(g, ncp = 30)
-
+g <- runPCA(g, feats_to_use = NULL)
+# feats_to_use = NULL since there are no HVFs calculated. Use all genes.
+screePlot(g, ncp = 30)
+
Figure 10.8: Plot of variance explained in the first 30 out of 100 principle components calculated
-
-
-
+
+
+
Figure 10.9: PCA plot showing the first 2 PCs (left), UMAP generated from first 15 PCs (right)
@@ -1047,37 +1047,37 @@
10.7.3 Dimension Reduction
10.7.4 Clustering
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 40
-)
-
-# takes roughly 1 min to run
-g <- doLeidenCluster(g)
-
-plotPCA_3D(g,
- cell_color = "leiden_clus",
- point_size = 1,
-)
-plotUMAP(g,
- cell_color = "leiden_clus",
- point_size = 0.1,
- point_shape = "no_border"
-)
-
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 40
+)
+
+# takes roughly 1 min to run
+g <- doLeidenCluster(g)
+
+plotPCA_3D(g,
+ cell_color = "leiden_clus",
+ point_size = 1,
+)
+plotUMAP(g,
+ cell_color = "leiden_clus",
+ point_size = 0.1,
+ point_shape = "no_border"
+)
+
Figure 10.10: 3D plot showing first PCs with leiden clustering annotations (left), UMAP plot showing leiden clustering results (right)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.11: Spatial plot with leiden clustering annotations.
@@ -1091,18 +1091,18 @@
10.8 Niche clustering
10.8.1 Spatial network
First a spatial network must be generated so that spatial relationships between cells can be understood.
-g <- createSpatialNetwork(g,
- method = "Delaunay"
-)
-
-spatPlot2D(g,
- point_shape = "no_border",
- show_network = TRUE,
- point_size = 0.1,
- point_alpha = 0.5,
- network_color = "grey"
-)
-
+g <- createSpatialNetwork(g,
+ method = "Delaunay"
+)
+
+spatPlot2D(g,
+ point_shape = "no_border",
+ show_network = TRUE,
+ point_size = 0.1,
+ point_alpha = 0.5,
+ network_color = "grey"
+)
+
Figure 10.12: Delaunay spatial network`
@@ -1113,65 +1113,65 @@
10.8.1 Spatial network10.8.2 Niche calculation
Calculate a proportion table for a cell metadata table for all the spatial neighbors of each cell. This means that with each cell established as the center of its local niche, the enrichment of each leiden cluster label is found for that local niche.
The results are stored as a new spatial enrichment entry called ‘leiden_niche’
-
+
10.8.3 kmeans clustering based on niche signature
-# retrieve the niche info
-prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
-# convert to matrix
-prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
-
-# perform kmeans clustering
-set.seed(1234) # make kmeans clustering reproducible
-prop_kmeans <- kmeans(
- x = prop_matrix,
- centers = 7, # controls how many clusters will be formed
- iter.max = 1000,
- nstart = 100
-)
-prop_kmeansDT = data.table::data.table(
- cell_ID = names(prop_kmeans$cluster),
- niche = prop_kmeans$cluster
-)
-
-# return kmeans clustering on niche to gobject
-g <- addCellMetadata(g,
- new_metadata = prop_kmeansDT,
- by_column = TRUE,
- column_cell_ID = "cell_ID"
-)
-
-# visualize niches
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_fill = "niche",
- # polygon_fill_code = getColors("Accent", 8),
- polygon_alpha = 1,
- polygon_fill_as_factor = TRUE
-)
-
-ggplot2::ggplot(
- cx, ggplot2::aes(fill = as.character(leiden_clus),
- y = 1,
- x = as.character(niche))) +
- ggplot2::geom_bar(position = "fill", stat = "identity") +
- ggplot2::scale_fill_manual(values = c(
- "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
- "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
- )
-
+# retrieve the niche info
+prop_table <- getSpatialEnrichment(g, name = "leiden_niche", output = "data.table")
+# convert to matrix
+prop_matrix <- GiottoUtils::dt_to_matrix(prop_table)
+
+# perform kmeans clustering
+set.seed(1234) # make kmeans clustering reproducible
+prop_kmeans <- kmeans(
+ x = prop_matrix,
+ centers = 7, # controls how many clusters will be formed
+ iter.max = 1000,
+ nstart = 100
+)
+prop_kmeansDT = data.table::data.table(
+ cell_ID = names(prop_kmeans$cluster),
+ niche = prop_kmeans$cluster
+)
+
+# return kmeans clustering on niche to gobject
+g <- addCellMetadata(g,
+ new_metadata = prop_kmeansDT,
+ by_column = TRUE,
+ column_cell_ID = "cell_ID"
+)
+
+# visualize niches
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_fill = "niche",
+ # polygon_fill_code = getColors("Accent", 8),
+ polygon_alpha = 1,
+ polygon_fill_as_factor = TRUE
+)
+
+ggplot2::ggplot(
+ cx, ggplot2::aes(fill = as.character(leiden_clus),
+ y = 1,
+ x = as.character(niche))) +
+ ggplot2::geom_bar(position = "fill", stat = "identity") +
+ ggplot2::scale_fill_manual(values = c(
+ "#E7298A", "#FFED6F", "#80B1D3", "#E41A1C", "#377EB8", "#A65628",
+ "#4DAF4A", "#D9D9D9", "#FF7F00", "#BC80BD", "#666666", "#B3DE69")
+ )
+
Figure 10.13: Leiden annotation-based spatial niches
-
+
Figure 10.14: Stacked barplot of leiden annotation composition by niche
@@ -1182,57 +1182,57 @@
10.8.3 kmeans clustering based on
10.9 Cell proximity enrichment
Using a spatial network, determine if there is an enrichment or depletion between annotation types by calculating the observed over the expected frequency of interactions.
-# uses a lot of memory
-leiden_prox <- cellProximityEnrichment(g,
- cluster_column = "leiden_clus",
- spatial_network_name = "Delaunay_network",
- adjust_method = "fdr",
- number_of_simulations = 2000
-)
-
-cellProximityBarplot(g,
- CPscore = leiden_prox,
- min_orig_ints = 5, # minimum original cell-cell interactions
- min_sim_ints = 5 # minimum simulated cell-cell interactions
-)
-
+# uses a lot of memory
+leiden_prox <- cellProximityEnrichment(g,
+ cluster_column = "leiden_clus",
+ spatial_network_name = "Delaunay_network",
+ adjust_method = "fdr",
+ number_of_simulations = 2000
+)
+
+cellProximityBarplot(g,
+ CPscore = leiden_prox,
+ min_orig_ints = 5, # minimum original cell-cell interactions
+ min_sim_ints = 5 # minimum simulated cell-cell interactions
+)
+
Figure 10.15: Cell-cell interaction enrichments and depletions (left). Number of interactions of each type found (right)
Most enrichments are self-self interactions, which is expected. However, 6–8 and 2–9 stand out as being hetero interactions that are enriched with a large number of interactions. We can take a closer look by plotting these annotation pairs with colors that stand out.
-# set up colors
-other_cell_color <- rep("grey", 12)
-int_6_8 <- int_2_9 <- other_cell_color
-int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
-int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
-
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_6_8,
- polygon_line_size = 0.1,
- polygon_alpha = 1,
- show_image = TRUE,
- image_name = "HE"
-)
-spatInSituPlotPoints(g,
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- polygon_fill_code = int_2_9,
- polygon_line_size = 0.1,
- show_image = TRUE,
- polygon_alpha = 1,
- image_name = "HE"
-)
-
+# set up colors
+other_cell_color <- rep("grey", 12)
+int_6_8 <- int_2_9 <- other_cell_color
+int_6_8[c(6, 8)] <- c("orange", "cornflowerblue")
+int_2_9[c(2, 9)] <- c("orange", "cornflowerblue")
+
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_6_8,
+ polygon_line_size = 0.1,
+ polygon_alpha = 1,
+ show_image = TRUE,
+ image_name = "HE"
+)
+spatInSituPlotPoints(g,
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ polygon_fill_code = int_2_9,
+ polygon_line_size = 0.1,
+ show_image = TRUE,
+ polygon_alpha = 1,
+ image_name = "HE"
+)
+
Figure 10.16: Spatial plot of enriched leiden annotation 6 to 8 interactions
-
+
Figure 10.17: Spatial plot of enriched leiden annotation 2 to 9 interactions
@@ -1245,18 +1245,18 @@
10.10 Pseudovisiumpvis <- makePseudoVisium(
- extent = ext(g,
- prefer = c("polygon", "points"),
- all_data = TRUE # this is the default
- ),
- micron_size = 1
-)
-g <- setGiotto(g, pvis)
-g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
-
-plot(pvis)
-
+pvis <- makePseudoVisium(
+ extent = ext(g,
+ prefer = c("polygon", "points"),
+ all_data = TRUE # this is the default
+ ),
+ micron_size = 1
+)
+g <- setGiotto(g, pvis)
+g <- addSpatialCentroidLocations(g, poly_info = "pseudo_visium")
+
+plot(pvis)
+
Figure 10.18: Pseudovisium spot geometries generated by makePseudoVisium()
@@ -1265,65 +1265,65 @@
10.10 Pseudovisium
10.10.1 Pseudovisium aggregation and workflow
Make ‘pseudo_visium’ the new default spatial unit then proceed with aggregation and usual aggregate workflow
-activeSpatUnit(g) <- "pseudo_visium"
-
-g <- calculateOverlap(g,
- spatial_info = "pseudo_visium",
- feat_info = "rna"
-)
-g <- overlapToMatrix(g)
-
-g <- filterGiotto(g,
- expression_threshold = 1,
- feat_det_in_min_cells = 1,
- min_det_feats_per_cell = 100
-)
-
-g <- normalizeGiotto(g)
-g <- addStatistics(g)
-
-spatInSituPlotPoints(g,
- show_image = TRUE,
- image_name = "HE",
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "total_expr",
- polygon_fill_gradient_style = "sequential"
-)
-
+activeSpatUnit(g) <- "pseudo_visium"
+
+g <- calculateOverlap(g,
+ spatial_info = "pseudo_visium",
+ feat_info = "rna"
+)
+g <- overlapToMatrix(g)
+
+g <- filterGiotto(g,
+ expression_threshold = 1,
+ feat_det_in_min_cells = 1,
+ min_det_feats_per_cell = 100
+)
+
+g <- normalizeGiotto(g)
+g <- addStatistics(g)
+
+spatInSituPlotPoints(g,
+ show_image = TRUE,
+ image_name = "HE",
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "total_expr",
+ polygon_fill_gradient_style = "sequential"
+)
+
Figure 10.19: Pseudo visium total detections per spot
-g <- runPCA(g, feats_to_use = NULL)
-g <- runUMAP(g,
- dimensions_to_use = seq(15),
- n_neighbors = 15
-)
-
-g <- createNearestNetwork(g,
- dimensions_to_use = seq(15),
- k = 15
-)
-
-g <- doLeidenCluster(g, resolution = 1.5)
-
-# plots
-plotPCA(g, cell_color = "leiden_clus", point_size = 2)
-plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE
-)
-spatInSituPlotPoints(g,
- polygon_feat_type = "pseudo_visium",
- polygon_fill = "leiden_clus",
- polygon_fill_as_factor = TRUE,
- show_image = TRUE,
- image_name = "HE"
-)
-
+g <- runPCA(g, feats_to_use = NULL)
+g <- runUMAP(g,
+ dimensions_to_use = seq(15),
+ n_neighbors = 15
+)
+
+g <- createNearestNetwork(g,
+ dimensions_to_use = seq(15),
+ k = 15
+)
+
+g <- doLeidenCluster(g, resolution = 1.5)
+
+# plots
+plotPCA(g, cell_color = "leiden_clus", point_size = 2)
+plotUMAP(g, cell_color = "leiden_clus", point_size = 2)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE
+)
+spatInSituPlotPoints(g,
+ polygon_feat_type = "pseudo_visium",
+ polygon_fill = "leiden_clus",
+ polygon_fill_as_factor = TRUE,
+ show_image = TRUE,
+ image_name = "HE"
+)
+
Figure 10.20: Leiden clustering in PCA (top left) and UMAP (top right) spaces, and in spatial plot with no image (bottom left), and with image (bottom right)